EP1123209A1 - Vorrichtung und verfahren zum registrieren von registermarken - Google Patents
Vorrichtung und verfahren zum registrieren von registermarkenInfo
- Publication number
- EP1123209A1 EP1123209A1 EP99971761A EP99971761A EP1123209A1 EP 1123209 A1 EP1123209 A1 EP 1123209A1 EP 99971761 A EP99971761 A EP 99971761A EP 99971761 A EP99971761 A EP 99971761A EP 1123209 A1 EP1123209 A1 EP 1123209A1
- Authority
- EP
- European Patent Office
- Prior art keywords
- web
- color
- marks
- control process
- mark recognition
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Withdrawn
Links
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B41—PRINTING; LINING MACHINES; TYPEWRITERS; STAMPS
- B41F—PRINTING MACHINES OR PRESSES
- B41F33/00—Indicating, counting, warning, control or safety devices
- B41F33/0081—Devices for scanning register marks
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B41—PRINTING; LINING MACHINES; TYPEWRITERS; STAMPS
- B41P—INDEXING SCHEME RELATING TO PRINTING, LINING MACHINES, TYPEWRITERS, AND TO STAMPS
- B41P2233/00—Arrangements for the operation of printing presses
- B41P2233/50—Marks on printed material
- B41P2233/52—Marks on printed material for registering
Definitions
- Zoom Lens Calibration present a very versatile multiprocessing system that performs a number of functions using the same hardware with additional software for each function.
- This multi-functional capability provides an attractive overall cost structure when compared with the cost of a number of individual and separate products to provide the same performance.
- the cost of the complete system can be considerably more than the cost of one or two separate systems.
- This disclosure describes a system that provides some of the benefits of Zoom Lens Calibration at a greatly reduced cost.
- the present invention describes new and improved features that provide significant new capabilities over those described in Zoom Lens Calibration.
- zoom Lens Calibration a method for obtaining initial register with random insertion of printing cylinders was disclosed. The accuracy of this method is more than sufficient to align all of the marks in their relative positions with no overlap. The software would then recognize each of the marks, calculate their position errors relative to their ideal position, and introduce corrections to align all marks to their ideal position.
- This invention relates to registration system for operations to be performed on a moving web, typically for multicolor printing.
- This system concentrates on the combination of analog and digital techniques to reduce measurement of error in web registration.
- the system concentrates on the use of photocells and conventional fixed-lens systems of image inspection to perform the registration function.
- the general registration system disclosed in this patent is the basis for many of the following patents mentioned in this application.
- This patent describes a photographic packaging system including a monitoring system which permits the operator to compares frames of film negatives with prints which are being cut so as to ensure that the proper prints will be packaged with the corresponding film negatives.
- the optics system described here is intended for manual quality control of the production process.
- This patent describes a method and apparatus for positioning an object in space in such a manner as to permit duplication of the positioning in a precise position using synthesized points on a television screen. While this patent does deal with distance measurements, it is a fundamentally different task to position an object in space and to determine the object's position relative to other objects, as must be done with the web printing process during the initial registration procedure.
- This patent describes the use of a TV camera to view the material and then electronically analyze the video image for alternating light and dark areas within a given length. This technique essentially replaces the use of photocells in the 2,802,666 Crosfield patent with the updated technology of television video. Note that no signal processing techniques are used in this patent save for crude edge detection algorithms.
- SEMICONDUCTOR DEVICES DURING MANUFACTURE This patent describes a system for aligning successive configurations of minute semiconductors during manufacture, the configurations being carried on a table which is under automatic control of a standard TV camera, pattern-recognition and motor control circuitry that corrects for TV camera geometrical and shading distortions, and a monitor which may be used by an operator to supervise the system.
- this patent is primarily concerned with determining the orientation of a given item of manufacture as it is processed, rather than determining the relative position of two marks within a given product of manufacture as is the case in a web printing process.
- the major difference here is that individual alignment relative to other manufactured product is irrelevent in the 4,233,625 patent, whereas in web printing applications it is the primary focus of interest .
- the web printing process is fundamentally different than the manufacture or positioning of piece parts, in that the initial register required by web print processes is a close manufacturing loop in which final web product material is manufactured according to data obtained from prior manufactured web material.
- This system permits comparison of a television image and the desired web image to be made and then adjustments can be made to the web printing process to compensate for the detected differences. While this is a closed loop system as describe in the disclosed exemplary embodiments in this application, the Wales patent still requires that a reference be provided for comparison with the television imaging system. The disclosed method and apparatus in this application dispenses with this requirement and permits initial register to be performed without the requirement of comparison web material. This has a significant time and cost savings for manufacturing, as the generating of an acceptable comparison web sample can take time and waste material. Note the following issues concerning this patent: (A) This is the only combination television-strobe patent dealing with web manufacturing.
- the Wales patent requires "proof” sheets, or target comparison material to be generated in order to perform proper comparisons and generate the required feedback control for the web printing operation.
- the Gnuttlel patent like the Wales 4,736,680 patent, describes a closed-loop registration control system in which optical data is sampled by means of a strobe and processed to control the overall production of the printed web material.
- the Gnuchtel patent does illustrate a moveable camera inspection station.
- the Gnuchtel patent also fails to allow the use of wide field of view lenses or Zoom Lenses to provide the ability to locate web registration marks over a wide field of view. This capability requires a more sophisticated approach to the calibration of the actual camera position with respect to the web material and is disclosed later in this document.
- the Gnuchtel is a first order manual solution to a much more complex problem of automatic initial register control.
- the Gnuchtel patent essentially brings new technology to the Wales patent but goes no further in improving the overall accuracy of the camera positioning techniques disclosed by Wales.
- the present invention and associated embodiments permit multitasking of functions described in the Gnuchtel and Wales patent using multiple field views over the entire width of the printed web and using the same inspection device (lens / camera / processing unit) .
- the teachings of the present invention go far beyond that in the Gnuchtel patent in that the disclosed Zoom Calibration method may be utilized to implement in a multitasking fashion any number of features in the Gnuchtel patent in combination with other web inspection/control functions which are not possible with the Gnuchtel/Wales technology.
- successive line scans generated by the scanner provide the equivalent of a two-dimensional raster scan of a strip of the web centered on the expected center line of registraton marks produced by the respective printing units.
- the centers of the respective registration marks are determined, and the cyclical (rotational) and traverse (lateral) positions of the printing unit adjusted in accordance with deviations from expected relative positions.
- the Sainio patent specifically describes the use of symmetrical, right-angle diamonds for facilitating the calculation of the center of the registration marks.
- Sainio only claims the use of photo-optical line scanner sensors (using a tungsten- halogen lamp as the illumination source) and does not make use of any Zoom Lens technology to obtain wide field registration or provide any method of overcoming the need for multiple cameras to perform wide-field registration operations.
- Sainio' s method and disclosure are limited to sensor arrays which are positioned perpendicular to the direction of web travel, meaning that the method described can only be used when the press is operating. The reason behind this requirement is that the sensor array must see a registration mark pass its position to enable the edge detection algorithms described by Sainio to operate properly.
- Sainio' s specific mention of diamond-shaped registration marks is designed in fact to aid this process and provide a method of overcoming inherent deficiencies in the image sensing method disclosed in this patent.
- the fact that Sainio' s method does not permit a wide field of view with accurate distancing restricts the use of this apparatus to low- resolution registration applications.
- To accurately perform high-quality print registration requires a variable field of view which not possible with the Sainio invention.
- Press web jitter can be defined as variations in the web velocity as it passes through the press due to mechanical characteristics of the press and resonating interactions between the web material and press mechanics as the web progresses through the manufacturing process.
- This jitter makes sensing as described in the Sainio patent troublesome, as it becomes difficult to accurately determine the exact relative position of the registration mark center because of this inconsistancy in web velocity. For example, if the web velocity suddenly increases during traversal past the sensing device, then the resulting image will generate a signal signature which is different than a normal registration mark.
- the presently disclosed invention makes use of stroboscopic synchronization of the image capturing device, and as such does not require that the press be moving.
- the use of image processing techinques in conjunction with the Zoom Calibration method disclosed herein permits a wide variety of registration marks to be used with no loss of accurate distance calculation measurement control.
- Another significant difference in the presently disclosed invention as compared to the Saino disclosure is that with image processing it is possible to achieve registration with arbitrarily configured registration marks.
- Image processing can, when coupled with the Zoom Calibration method, permit the use of patterns within the web product to be used as registration marks for distance calculations . No prior art system claims to permit this type of registration using the final product itself as the registration mark.
- the stroboscopic synchronization of the presently disclosed invention combined with image processing techniques and accurate distancing provided by the Zoom Calibration method permit press jitter to be compensated for in the web inspection/control function.
- Brunetti patent is the lack of any support for wide-field Zoom Lenses which permit the location of widely disparate registration marks. Furthermore, Brunetti neither claims nor discloses any method to calibrate the accurate position of a camera or image capture device in respect to the web material registration marks. No method is provided to compensate for mechanical wear in the system as well as permit compensation for registration drift across a wide width web.
- the Brunetti patent address only a single-function inspection system.
- the disclosed Zoom Calibration method permits a multi-function capability within the same optical camera system. This would not be possible with the Brunetti teaching, as it discloses no method by which wide-field variations may be captured, nor is there any camera distance calibration method to permit accurate distance measurements across the web width.
- the Zoom Lens Calibration patent by Lewis, et. al. presents a very versatile multiprocessing system that performs a number of functions using the same hardware with additional software for each function.
- This multifunctional capability provides an attractive overall cost structure when compared with the cost of a number of individual and separate products to provide the same performance. If, however, only one or two of the functions are required, the cost of the complete system can be considerably more than the cost of one or two separate systems .
- This disclosure describes a system that provides some of the benefits of pending application Zoom Lens Calibration at a greatly reduced cost. In addition it describes new and improved features that provide significant new capabilities over those described in Zoom Lens Calibration.
- New applications that utilize these unique features provide significant reductions in waste material as will be explained in this disclosure. Particularly affected are major areas of full four-color printing in direct mail, newspaper, and commercial printing.
- a new method for rapidly achieving initial register is disclosed using fixed lens where initial register errors can be significantly greater than the spacing of the marks. This overcomes the limitation of all current register controls that require manual adjustment of the marks until all marks can be identified by the software before automatic initial register can be performed. This feature and a method for viewing faint or very light marks is described which provide for substantial reductions of waste material during initial register and during normal register control.
- Automatic calibration using duplicate marks described in "Zoom Lens Calibration" enable a host of new automatic register control applications hereto fore not possible.
- the camera can be mounted where continuous focal distance changes, and variations in web velocity occur such as automatic register control on a blanket to blanket web offset presses with top and bottom staggered printing units.
- Automatic register control of colors and cut to print can now be accomplished with the camera no longer synchronized with the printing with the camera-strobe mounted on a shingle delivery of a cutting and/or creasing press and/or a sheeter.
- the traverses, while motorized, are essentially manually positioned by an operator over the printed web during press operation. Lateral adjustments in this context are always accomplished via manual operator control. Note that the narrow field of view in both these implementations (inherent in any fixed magnification lens system) restricts the ability of the system to compensate for lateral web shift as the press heats up or the web material shifts during manufacture.
- the objects of the present invention are (among others) to circumvent the deficiencies in the prior art and affect the following objectives:
- zoom Lens Calibration the camera, Zoom Lens and strobe were mounted on a movable motorized traversing mechanism. This provided the capability for obtaining images at any position on the entire web width and repeat range. Thus, 100% of the repeat length and web width could be inspected at different magnifications depending upon the zoom ratio of the Zoom Lens.
- additional features were provided such as color-to-color, print-to-punch, and ront-to-back register control. Almost all of these features are required for applications where the raw material is a roll that is unwound, printed, and/or converted in some manner and then rewound in a finished roll.
- the raw material is a roll that is unwound, printed, and/or converted in some manner into finished products instead of a roll.
- a sample of the finished product is easily obtained from the delivery of the machine thus eliminating the need for traversing cameras.
- fixed lens cameras can be mounted on a motorized traverse providing the capability of performing a number of measurement and control functions including viewing of magnified images.
- the technology of lens calibration as disclosed in Zoom Lens Calibration is required when using a fixed lens to eliminate a manual calibration of the lens field of view and to automatically compensate for focal distance changes when the camera is moved to different positions across the web.
- the duplicate marks are also important for recognition purposes in providing automatic initial register.
- the duplicate marks are instrumental in the automatic detection of mark patterns enabling very rapid setup of the system reducing material waste and loss press time.
- the present invention includes a means for enhancing light colors with very fast image processing by restricting the search areas where the light marks should be located and enhancement of the surrounding areas.
- Mark spacing (distance between adjacent marks) was originally selected so as to provide for non-overlapping marks due to errors in the making and the mounting of the printing plates. Sufficient distance between adjacent marks in Zoom Lens Calibration allowed the plates to be mounted with adjacent marks in their respective positions without overlap. This technique enables the use of the automatic initial register function immediately without the need for the operator to first manually separate marks that are overlapping. In reality other factors such as the lack of a zeroing function for centering the mechanisms used to adjust both circumferential and lateral register produce initial register errors that are far greater than the plate errors. The magnitude of these errors vary from machine to machine and are not predictable. During actual running conditions the mark spacing could be very small which is desirable as it allows the hiding of the marks in the artwork where they cannot be seen.
- the present invention teaches a method where very small mark spacing may be used suitable for normal running conditions.
- a very rapid means is provided for initially aligning the marks so that they do not overlap in their area of recognition thus facilitating automatic final register.
- the advantage of recognizing objects is automatic register control can be applied in totally new applications where it was not possible in the past. This includes all applications other than those where both objects were a printed mark of a specific size.
- Providing the images for both objects are in the same image captured by the camera, accurate distance measurements can be made using the teachings of Zoom Lens Calibration. These measurements can be used for adjusting motors providing automatic register control.
- the camera can be mounted at any location where both images can be viewed independent of the distance from the correction mechanisms.
- the camera need not be synchronized with the printing process and images can be taken at random for purposed of measuring color registration form a mark pattern as well as print to cut accuracy in both the X and Y directions.
- the design of the system provides great diversity on the number of individual cameras that can be used. They can be a combination of color or black and white depending upon the application.
- the cameras can be illuminated with pulsed or Xenon flash tubes, pulsed infrared, and/or near infrared light sources and with pulsed ultraviolet light sources to provide for a number of different applications. These applications include conventional color-to-color register control, object-to-mark register control, reading of invisible marks such as used in security printing, automatic inspection of coatings for voids. Any number of cameras can be in operation at the same time using the same hardware with different software.
- a single monitor may be used which provides both operator control through a touch screen and provides viewing images that are selected specifically for viewing register marks and any other point of interest which can be viewed using a fixed lens.
- Software enables the adjustment and operation for all cameras through the touch screen with an automatic alert whenever operator attention is required with an image display indicating the area of concern.
- Software zoom capability provides enlargement of marks and spacing allowing very accurate identification of mark position errors for obtaining initial registration of marks or objects.
- the remote communication capability over telephone lines or other communications means provides a whole new means for interacting with equipment and customers. Some of these new functions include diagnosis of hardware and software, the capture of complete images for research and development, record press performance operation such as register capability, provide documentation for the customer of any parameter measured and controlled. Pre-Registration Process
- the image based pre-registration process provides for a means to complete the initial registration and fine registration processes in a timely fashion, saving time and material waste.
- a touchscreen monitor and user interface computer system is used to display screens, menus, and images. Images are displayed on the touchscreen monitor such that the user can "touch" a point on the image display to locate a specific object or mark being printed.
- the objects, or "targets”, located may be any object provided all the objects located adhere to the target method used. Target methods will be described in detail later in the document.
- a common point on all of the objects would be identified. For example, the user would touch the intersection of crosshairs on all crosshair objects, or the upper left corner of a box on all box objects.
- the point selected on the object is not relevant. However, consistency of the point selected on the object is required from target to target.
- global X and Y coordinates of the object may be assigned to the locations of the object on the images.
- the object coordinates are compared to designated key object coordinates to determine the relative distances the objects are apart.
- offsets are then used to determine individual station motor movements. Commands are then issued to move all station motors the appropriate offset distance. When motors are finished, all stations should be roughly in register. From the rough register state, the fine register control system can automatically recognize the mark pattern. At this point the fine register control system holds the press in register.
- One application of this process is the color-to-color initialization. By selecting targets from stations printing on the same side of the web, the offsets can be computed, and motors moved to bring all stations on the same side of the web into rough register.
- Color-to-Bindery Process Another application is a color-to-bindery process. For example, if a printing job requires lineholes, the color stations may be brought into rough register with the linehole bindery operation. An actual linehole is targeted and set as the key. Then, the color stations are targeted. The target for a color station typically is a linehole bug in this case. A linehole bug is a hollow box shape, which when in register, superimposes the linehole. All color stations are moved into rough initial register to the linehole . Front-to-Back Color-to-Color Initialization
- Another application is the front-to-back color-to-color initialization.
- the color stations of one side of the web can be brought into rough register with the key color on the opposite side of the web. This is possible thru the setting of global coordinates. Global coordinates will be described in detail later in the document.
- Register errors can be accurately measured for applications where the web speed is not constant which heretofore was not possible with existing technology.
- An example will be disclosed where this feature enables a substantial reduction of waste material.
- Zoom Lens Calibration enables applications where the camera can be located on an unsupported web where web flutter will cause erratic changes in focal distance. For every image the distance measurement is recalibrated. Thus, these variations have no affect on overall accuracy.
- This unique feature enables new print registration applications such as the location of a camera directly after a print unit in web offset printing where substantial web flutter is present with the result of a substantial reduction in waste material.
- the superimposed targets method uses printed control objects as the targets which, when in final register, are superimposed print on the web.
- the printing press is in registration, all of the printing stations would print an object such that all of the objects align on top of each other on the web.
- the X and Y coordinates of the objects would be the same.
- a printed control object is an ink deposit of any shape by a printing station.
- Typical control objects are, but not limited to: cross hairs, rectangles, squares, circles, horizontal lines, vertical lines, arrow shapes, star shapes, or any combination thereof.
- the control object is an ink deposit that is used as an aid by the press operator to bring the press into (and hold) final registration.
- the control object is not typically an object of the final product, rather an additional object either within, among, or adjacent to the final product.
- the use of a control object, or "bug" is a common practice in the industry.
- control object is used extensively, this method does not require a control object. Any objects that are printed superimposed when in register would suffice.
- the present invention utilizes control objects as targets that currently exist in the manual initial register process. Press operators and the art department that generate cylinder plates are familiar with the use of the object .
- the present invention is prone to error where the operator may select a target that may not be superimposed with the key when the press is in final register. In this case the erroneous target station will be moved significantly out-of-register .
- the mark pattern alignment method uses the marks of a mark pattern as the target control objects. Each mark of the mark pattern is located on the web and corresponding X and Y coordinates are stored with the mark. The mark is then identified by which printing station prints the selected mark. The current mark pattern of the job is then selected for the camera web side. The actual in register locations of the stations are defined in the mark pattern. The stations are moved such that the marks end up in the alignment and orientation of the mark pattern.
- the present invention utilizes one set of control objects for initial and final register processes. This would be significant in printing work where the majority of the web is reserved for final product. Disadvantages
- the present invention requires another control object area of the web where cylinders cannot print.
- the drag target method utilizes the touchscreen interface in that the user points to the object on the image, and "drags" the cursor to the desired in register location on the image.
- the current location of the cursor defines the desired location of the object. Coordinates are set, and offsets can be computed. Motor movement is started, and the user watches the object move into register in the updating image on the touchscreen.
- the present invention simplifies the process from a user interface point of view.
- FIG. 1 illustrates an exemplary Press Configuration with Zoom Lens
- FIG. 2 illustrates an exemplary detailed Two-Camera Zoom Lens System
- FIG. 3 illustrates an exemplary Press Configuration Simplified System
- FIG. 4 illustrates an exemplary Two-Camera Simplified System
- FIG. 5 illustrates an exemplary Two-Camera Simplified System with traverse
- FIG. 6 illustrates a Mark Pattern with Four-color Duplicate Marks
- FIG. 7 illustrates an exemplary Mark Pattern with Seven-Color Duplicate Marks
- FIG. 8 illustrates an exemplary Mark Pattern with Nine- Color Duplicate Marks
- FIG. 9 illustrates exemplary mark positions after final alignment
- FIG. 10 illustrates exemplary register setup mark positions
- FIG. 11 illustrates exemplary register setup duplicate marks
- FIG. 12 illustrates exemplary register setup with all marks
- FIG. 13 illustrates exemplary typical mark positions after changing plates before initialization alignment
- FIG. 14 illustrates exemplary select print stations cross hair within a search display
- FIG. 15 illustrates exemplary assignment of print units within a search display
- FIG. 16 illustrates exemplary initial register required corrections for alignment
- FIG. 17 illustrates an exemplary drag method for rough register alignment
- FIG. 18 illustrates Single monitor with multiple camera display and object registration
- FIG. 19 illustrates waste reduction statistics for an exemplary newspaper color registration application
- FIG. 20 illustrates an exemplary newspaper press configuration
- FIG. 21 illustrates an exemplary tension-velocity change due to blanket stagger
- FIG. 22 illustrates an exemplary tension-velocity display graph
- FIG. 23 illustrates an exemplary heat set printing press
- FIG. 24 illustrates magnified heat set printing press with focal distance changes
- FIG. 25 illustrates field of view change with change in focal distance
- FIG. 26 illustrates mark patterns from the operator and gear side view
- FIG. 27 illustrates aligned mark pattern from the operator side view
- FIG. 28 illustrates XY correction coordinates of mark patterns
- FIG. 29 illustrates automatic register alignment in two dimensions
- FIG. 30 illustrates an exemplary system process flowchart implementing an image based pre-registration system
- FIG. 31 illustrates an exemplary system process flowchart implementing an overall mark recognition system
- FIG. 32 through FIG. 56 illustrate exemplary system process flowcharts implementing the details of an overall mark recognition system
- FIG. 57 illustrates an exemplary system process flowchart implementing the basic functions of a registration mark recognition system
- FIG. 58 through FIG. 59 illustrate exemplary process flowcharts implementing the basic functions of a registration mark recognition method
- FIG. 60 through FIG. 76 illustrate exemplary system process flowcharts implementing the details of a light mark detection system
- FIG. 77 illustrates an exemplary system process flowchart implementing a color monitoring system
- FIG. 78 illustrates an exemplary system process flowchart implementing total quality control and waste reduction system for a typical web offset printing application
- FIG. 79 illustrates an exemplary system process flowchart implementing a two dimensional optimization and/or correction of all registration variables
- FIG. 80 illustrates an exemplary newspaper waste recovery profile based on a typical product mix possible using an exemplary preferred embodiment of the present invention
- FIG. 81 illustrates an exemplary variant of the present invention in which a local print registration system is remotely controlled and/or updated via a communications network and remote processing system with associated software database, thus permitting remote diagnosis and/or control of an operational printing press system;
- FIG. 82 illustrates an exemplary system process flowchart implementing a color monitoring and/or control variant of the present invention wherein color ink keys are modulated to provide absolute color control;
- FIG. 83 illustrates an exemplary system process flowchart implementing a color monitoring and/or control variant of the present invention wherein ink/water balance is modulated to provide absolute color control.
- the present invention may be aptly described in terms of exemplary system block diagrams and procedural flowcharts. While these items are sufficient to instruct one of ordinary skill in the art the teachings of the present invention, they should not be strictly construed as limiting the scope of the present invention.
- system block diagrams may be combined and rearranged with no loss of generality, and procedural steps may be added or subtracted, and rearranged in order to achieve the same effect with no loss of teaching generality.
- the present invention as depicted in the attached exemplary system block diagrams and procedural flowcharts is for teaching purposes only and may be reworked by one skilled in the art depending on the intended target application.
- PC personal computer
- the term 'personal computer' should be given a broad meaning in this regard, as in general any computing device may be utilized to implement the teachings of the present invention, and the scope of the invention is not limited just to personal computer applications, or to a specific computer processor.
- touch screen interface as the primary means of communicating to the press operator. While this is the preferred method of communication, the present invention is in no way limited to this means of communication. Thus, the term 'touch screen' and/or operator interface should be construed in its broadest sense as being any means of communication between an operator (either locally or remotely) and the other components of the present invention embodiment.
- any geometry may be suitable for use in the disclosed system. Therefore, the present invention specifically anticipates the use of common geometries such as rectangles, squares, circles, regular polygons, etc. within a wide variety of implementations utilizing the teachings of the present invention.
- duplicate marks mentioned herein can be in any configuration, including but not limited to rectangular arrays and other geometric patterns. In many preferred embodiments, however, columnar and rectangular arrangements have advantageous properties with respect to processing the web image data.
- the present invention may be advantageously applied to offset printing presses, nothing in the teachings of the present invention limits the scope to this particular application.
- one of the strong advantages of the present invention is that it may be retrofitted or integrated into a wide variety of older printing presses to improve their overall performance and waste generation characteristics.
- the present invention may be applied advantageously to both new and old presses in either an integrated or after-market configuration.
- Station - a set of cylinders on a printing press that deposits one color of ink.
- Targets any printed object used to define the location of the ink deposits of a station.
- Mark Pattern - a set of marks printed to with a predefined orientation and dimensions.
- a web material (5701) incorporates register marks (5702) that are inspected by an image acquisition system (5703) .
- This image is manipulated by an image processor (5704) that then interacts with an operator display (5705) and press motor controls (5706) to affect print registration of the web material (5701) .
- a web preregistration step (5801) initially aligns the web registration.
- a mark registration step (5802) that may include a light mark registration step (5803) .
- the output of these steps is optionally displayed on an operator display (5805) .
- the web press may be adjusted (5804) to achieve the desired print registration.
- the process of FIG. 58 may be optionally augmented by steps that move the image acquisition camera or select a new image acquisition camera
- FIG. 1 comprising four printing stations (101, 102, 103, 104) that print four colors on both sides of the web (116) .
- the web starts from the unwind stand (100) progresses through the four printing stations through the two-camera traversing mechanism (107) to either an unwind (108) or to a sheeter (109) .
- the two-camera traversing unit is the same as shown in Zoom Lens Calibration and is shown in detail in FIG. 2.
- the two cameras (201, 202) are equipped with Zoom Lens (203, 204) respectively and are mounted on traversing mechanisms (207, 210) .
- the system has the capability of stopping the motion, and freezing the image on the monitor so that detailed inspections can be made.
- the system can also be used to scan a mark pattern and automatically maintain register.
- FIG. 3 shows the same printing press of FIG. 1 with a much simplified two-camera system (301) .
- FIG. 4 illustrates a detail of (301) showing the two camera assembles (401, 402) .
- These two cameras are used for automatic register control and any other function that can be accomplished with a fixed lens stationary camera. They can be permanently positioned over a section of the printed web that contains the mark pattern. Thus, they can use a fixed lens and can be manually positioned across the web, therefore eliminating the need and expense of a Zoom Lens and the traversing mechanisms .
- the viewing monitor (112) of FIG. 1 can be eliminated requiring only the touch screen (302) of FIG. 3.
- the touch screen is used in this disclosure for the dual purpose of providing real time images of the mark pattern and as an operator interface.
- the two stationary cameras with fixed lens can provide the same functions of automatic register control as that of FIG. 1 or any other function that can be performed with stationary cameras with a fixed lens.
- the cost of the system of FIG. 3 is less than one/third the cost of the system of FIG. 1.
- the stationary fixed lens cameras of FIG. 4 can be motorized, thus enabling movement of the cameras laterally across the web in the same manner as the cameras equipped with Zoom lens of FIG. 2.
- the motorized fixed lens cameras are shown in FIG. 5.
- the top camera (501) has a traversing mechanism (503) driven by motor (506) .
- the bottom camera (502) has a traversing mechanism (504) driven by motor (507) .
- two calibration plates (509, 510) are incorporated in the system (one for the top camera and one for the bottom camera) that allow the mounting and viewing of calibration plates to provide for absolute positioning and for absolute color measurement.
- precision distance measuring calibration plates were used to perform the same function as duplicate marks to calibrate any zoom position.
- These calibration plates can also contain color chips to calibrate colors (yellow, cyan, magenta, black, etc.) that would enable automatic initial positioning of keys further reducing make ready waste.
- This fixed lens two cameras traversing assembly is considerably smaller and less costly than the Zoom Lens two- camera system of FIG. 2. It provides the capability of performing a number of new functions described herein that further automate register control and in addition measure and control color.
- Material waste is composed almost totally of misregister and initially achieving color in the web offset process.
- the two-camera fixed lens system traverse of FIG. 5 is suited to address both of these areas of waste at very reasonable costs.
- the small size of the mechanism of FIG. 5 enables its application on most all existing web offset presses that are inline with a folder or sheeter where the Zoom Lens are not needed nor suitable because of their large size .
- a single touch screen monitor is also used for both viewing and operator control thus reducing costs even further than that of FIG. 2.
- the fixed lens traversing system of FIG. 5 is capable of performing all functions of the Two-Camera Zoom Lens System of FIG. 2 with the fixed lens chosen for one specific zoom ratio.
- the field of view is chosen for the higher resolutions required for automatic register control both in the X and Y directions for printed marks and objects and for monitoring color.
- the traverse module is considerably smaller than the Two-Camera Zoom Lens System of FIG. 2 and thus is easily mounted in strategic positions on presses to small for the larger system of FIG. 2.
- duplicate marks were used primarily for the purpose of calibration of the Zoom Lens.
- the duplicate marks provide automatic compensation for all other variables including errors in placement of the marks in prepress, on press camera focal distance changes, speed variations, and variations due to paper modulus and tension changes. Correction for all of these variables occur every time an image is taken and the duplicate marks are used in the manner as described in Zoom Lens Calibration and this disclosure.
- Duplicate marks are used for the same purpose with a fixed lens as minor variations in focal distances due to positioning of the camera across the web would create register errors.
- the duplicate marks enable automatic calibration of the fixed lens without the need for providing a very accurate mechanism for maintaining an accurate focal distance across the entire web width.
- the duplicate marks provide calibration of distance measurement for every image which provides the capability to mount a camera on unsupported web with no adverse affect on register accuracy in the presence of web flutter.
- An example is the mounting of cameras directly after the printing nip on a web offset press and where back up rollers cannot be added as they would damage the printed image as the ink is not dry at this point. The result is further reductions in waste material as corrections can be introduced much sooner.
- the duplicate marks serve another very important function in that they enable the identification of the mark pattern providing for a completely automatic setup of the register control from finding and locating the register mark pattern to automatic operation. Mark Patterns
- FIGs. 6-8 show three horizontal mark patterns which provide for a total of 4-, 7-, and 9-colors respectively.
- the duplicate marks (601, 602) illustrated are printed by the same color station.
- the duplicate marks (701, 702, 703, 704) are printed by the same color.
- the duplicate marks (801, 802, 803, 804) are printed by the same color station.
- the simplicity and uniqueness of the mark recognition system to be described eliminates the need to coordinate a specific mark position in the pattern with a specific printing unit to print that color. That is, the prepress department can select any mark position and its color without regard for the press configuration or its installed register controls.
- the operator (as will be explained) need only select the color unit that prints the color and its position within the selected mark pattern as will be described.
- the identification or recognition of marks based on their color with full color images from a color camera is critical in the process of rapid setup and initial register to be disclosed.
- FIG. 30 illustrates an exemplary embodiment of the
- FIG. 30 is a step-by-step process of this concept (3000) .
- the press Before the imaging process can be started, a few preliminary steps must be undertaken on the press. First the cylinder printing plates must be mounted on each station's printing cylinder (3001). Important web variables such as web width and web length must be entered into the system (3002) . All limited range register correction motors must have their gearboxes centered (3003) . This is a process of driving a register motor for the total range time in one direction and then running the same motor in the other direction at approximately one-half the total range time. At this point the press is brought up to some nominal printing speed and it is made to print the material (3004) .
- a touch screen interface may be used to locate all register targets on the printed material. This process is described in detail in the associated description for FIG. 14 contained herein. While a touch screen interface is utilized in many preferred embodiments of the present invention, any graphical interface may be suitable in this application.
- To locate the targets on the printed material the appropriate web (3005) and camera (3006) must first be selected on the touchscreen. The touchscreen displays an image of the current location of the camera
- the camera is moved to locate the first target (3008) .
- the desired target is touched, assigned a station number, and given coordinates as to its location on the web (3009) .
- This process is repeated with all desired targets (3010) .
- a desired register mark pattern and key station are selected (3011) . This process is repeated for each desired camera (3012) and web (3013) .
- FIG. 8 The mark pattern of FIG. 8 is used to describe the initial rough register alignment procedure.
- FIG. 8 includes dimensions of the mark pattern stored in memory to identify the mark positions and their offsets for automatic alignment as will be explained.
- FIG. 9 shows the mark pattern of FIG. 8 as it will appear after all of the marks are aligned according to the following procedure.
- Each different colored mark is represented by a different cross hatching for the purpose of this discussion.
- all marks can be easily identified on the monitor as they are of different colors.
- a high color fidelity camera and monitor are used with RGB color rendition thus allowing the identification of any color that is printed.
- the steps in achieving initial register are as follows:
- FIG. 10 illustrates the Register Setup screen that is selected from the main menu (not shown) .
- the four-by- three mark pattern is the same as mark pattern as FIG. 8. Any one of several programmed mark patterns can be selected by repeatedly pressing the button labeled "Mark Pattern" (1013) of FIG. 10 until the selected mark pattern appears.
- the next step is to assign the number of the printing unit to each mark position. This is accomplished from information contained in the color proof that is given to the press operator before he starts up the job.
- the color proof includes the mark pattern with the actual color of the mark in each position of the mark pattern. Each mark in the pattern is matched with the number of the printing unit that will print that color as follows.
- FIG. 12 shows the remaining mark positions selected in the same manner depressing each button in turn until the number of the printing unit that prints that color appears in each box.
- FIG. 12 shows mark (1205) selected as printing unit 5, mark (1206) selected as printing unit
- mark (1207) selected as printing unit 9 mark (1208) selected as printing unit 2
- FIG. 13 illustrates a typical image of all marks before initial register has been achieved as shown in FIG. 9. Note the mark position errors are due to plate register and mounting errors that can result in marks overlapping (1306, 1302), touching (1307, 1311), and very close (1304, 1310) . Every time new plates for a new job are installed an entirely non-predictable pattern will appear with any or all of these conditions present. Since the range of the register correction mechanisms is about plus or minus one-eighth of an inch, the initial register errors must fall within this range or it would not be possible to get the marks in register. 3. Calculation of initial register errors is accomplished as follows. The initial register pattern of FIG.
- FIG. 14 is shown in FIG. 14 on the touch screen monitor with a superimposed menu (1413) .
- the button (1414) "Define Targets” is pushed and the cross hair appears (1417) .
- the operator puts his finger on the intersection of the X and Y cross hair (1417) and drags it to the center of each mark in sequence.
- the "Move Motors" push button (1515) is pressed. All motors will move the exact amount required for the marks to take their correct position in the mark pattern.
- the correction motors are generally 2-phase synchronous motors that correct proportional to the time that they are energized. These ratios are entered and stored in the computer so exact correction can be introduced thus assuring the marks will be very accurately moved to their proper position. Note that the press need only run for as long as a single image can be obtained from which the initial register errors are determined according to the procedure just described.
- FIG. 16 is generated at the completion of step 3 and before the correction motors are energized. It displays the color units (1606) , the camera associated with these colors unit, the amount of side (1604) and running register (1605) errors that with correction will center all marks within the correct position in the mark pattern. It also displays the amount of allowable correction dictated by the range of the correction mechanisms for this example as 0.125-inch in the X direction (1602) and the Y direction (1603) .
- the drag method seems simpler as it reduces the number of steps in the initial register process. However, it is more prone to error as the operator must perform two operations on each mark selection instead of one. The procedure is more suited to dragging objects to areas where they will be in register with printed marks.
- An advantage of the web offset printing process is copy changes can be made quickly and very inexpensively more so than with any other printing process. Each time a copy change is made, new plates with the new copy must replace the old ones. On all presses when new printing plates are mounted on the printing cylinder, there are color-to-color register errors due to variations in plate making, errors in mounting the plates on the press and errors due to the mechanical mechanisms which correct register not being in their ideal center position. These errors are considerably greater than the normal register variations caused by variations in the printing substrate during printing. As an example on most of the newer single repeat perfecting web offset presses have color register variations due to variations in the printing substrate and will stay within about 0.005 to 0.010 inch without register controls and to about 0.001 or 0.002 inch with register controls.
- initial register techniques enable very small and closely spaced marks that provide automatic register control under normal running conditions with a means of rapidly aligning the marks when plates are changed and where register misalignment may be orders of magnitude greater than the register errors encountered during steady running conditions .
- the image captured by the present invention can be magnified both optically and electronically.
- the selection of the fixed lens field of view determines the optical magnification and is a function of the application.
- a resolution of 0.001 inch is desired, which for a 512 by 512 pixel array provides a field of view of about 0.5 inches. This is more than sufficient to accommodate the initial plate areas as has been discussed.
- FIG. 18 illustrates a composite image of the four fixed cameras with fixed lens of FIG. 3.
- Items (1804) and (1803) of FIG. 18 are displays of vertical mark patterns from the top and bottom cameras of (301) FIG. 3 which are shown in more detail as (401) and (402) of FIG. 4. These mark patterns are similar to the mark pattern of FIG. 6 but with the marks arranged in the vertical direction and for six colors.
- the mark or object pattern of (1801) is an image from (308) of FIG. 3 that includes a line hole object (1806) that is registered to mark (1805) .
- Software automatically recognizes the round object, calculates the center of the circle and measures the distance form the center of the circle to the center of the mark (1805) . Any variation is introduced as correction through a motorized differential attached to the line hole punch (309) of FIG. 3 (not shown) .
- Mark pattern (1802) is an image taken from camera (304) of FIG. 3 showing the cut (1807) of FIG. 18 that is (312) of FIG. 3 and mark (1808) that is one of the marks shown in FIG. 3 as (311) .
- the cut and the mark are recognized with the distance (1809) of FIG. 18 maintained constant. Correction is introduced to motorized web compensator (307) of FIG. 3. All of the advantages of initial register are available for objects as well as for marks.
- Camera (304) of FIG. 3 is shown mounted at the delivery of the sheeter where the edges of sheets (306) can be viewed.
- a mark of known dimensions or duplicate marks (312) are located near the cut edge (311) and are used to calculate the print to cut register errors in both X and Y. This information is used to provide automatic correction to a motorized web compensator (307) to maintain cut to print register. This is to maintain distance (310) constant.
- the camera is no longer synchronized to a continuously moving web.
- the accuracy is independent of web speed as well as focal distance. This feature is not available using the teachings of the prior art.
- This portion of the disclosure illustrates an exemplary application of automatic register control for multi-web newspaper printing presses that have at least one web of four-color process printing on one or both sides of the web.
- FIG. 19 illustrates a spreadsheet representing the waste reduction that can be achieved on a typical newspaper printing press that prints a number of webs in black and white with a single four-color web.
- This example is for a typical small newspaper with a circulation of 50,000 (1901) having approximately 36 pages of print (1913) .
- the actual waste for achieving initial register for each plate change is about 2000 newspapers. With a 90% or more reduction of this waste of 1000 newspapers (1914), a savings of 4.5 tons (1924) of newsprint per week or 235 tons of newsprint per year (1925) can be achieved. This represents a sizable reduction in newspaper waste that heretofore has been either recycled or deposited in landfill.
- FIG. 20 illustrates a typical eight-color unit newspaper printing press with five webs feeding into a folder and delivering finished newspapers.
- Five roll stands (2001, 2002, 2003, 2004, 2005) feed printed webs into the printing units that are combined at folder (2014) into finished newspapers.
- Roll stands (2002, 2003, 2004, 2005) feed into printing units (2009, 2008, 2007, 2006), respectively.
- Each of these printing units prints a single color (black) on both side of the web.
- Roll stand (2001) feeds into printing units (2010, 2011, 2012, 2013) .
- These four printing units print the four-process colors (yellow, magenta, cyan, and black) on both sides of the web.
- four colors are registered (both initially and during the run) manually by the operator, and as to date automatic register controls have not been successful in this application for a number of reasons to be explained herein.
- FIG. 20 illustrates a typical five-web newspaper printing machine with one four-color lead in a horizontal press configuration.
- the color lead enters printing unit (2013) and continues through printing units (2012, 2011, 2010) where it exits and runs vertically through the two-camera unit (2015) similar to that illustrated in FIG. 4 and then combines with the other black and white webs at the folder.
- the only position where the two cameras can be mounted is directly after the fourth printing unit where all four colors of a mark pattern printed on both sides of the web can be viewed.
- FIG. 21 illustrates a greatly magnified view of this area of FIG. 20. It shows the top and bottom cameras (2102, 2101) and printing unit (2111) that is printing unit (2010) illustrated in FIG. 20.
- This printing unit has top plate cylinder (2105) that transfers its image to blanket (2103) and bottom plate cylinder (2106) that transfers its image to blanket (2104) .
- the image of top plate cylinder (2105) and the image of bottom plate cylinder (2106) are transferred to the web (2112) .
- blanket cylinder (2103) is staggered in relationship to bottom blanket cylinder (2104) .
- the printing unit is typically designed in this manner so that the gaps in the plate cylinder where the plate is attached and locked into the printing cylinder do not occur at the same time as they would if the blanket cylinders were not staggered. This prevents a resonance effect due to gear bounce and vibration that causes the undesirable effect called doubling that appears in the print as a double image.
- the ink is tacky and the stagger of the blanket cylinder tends to have the web follow the leading blanket or top blanket (2103) shown as (2107) .
- the web will release as shown at (2108) .
- This causes a pulsating tension change with a velocity profile (2212) as illustrated in FIG. 22.
- This abrupt velocity change occurs for each revolution of the blanket cylinder (2103) of FIG. 21.
- the present invention use of a strobe in some embodiments freezes the entire image in about 5 microseconds that at 1000 feet per minute is less than 0.001 inch of movement. All measurements are relative to each other and this disturbance has no affect on the accuracy of the measurement.
- the present invention configuration permits webs that are unsupported by press machinery to be inspected and registered without the need for any direct contact to the web by any stabilizing machinery. This is a direct result of the fact that focal length changes between the camera/sensor device and the web can be fully compensated for using image processing techniques.
- photocell and linear array systems require an absolutely constant web speed to provide accuracy. Any speed variation during the time that the marks pass the web scanner would produce erroneous error calculations in these prior art systems.
- this limitation as well as the lack of an initial register system as disclosed precludes the use of the photocell and linear array in applications where erratic speed changes are present such as just described.
- the present invention may be placed at any point in the web processing line, including after the last print station at any position alongside any position in which the web contains ink that is not dry, after any staggered plate, top/bottom cylinder, or the like, as well as after any top/bottom blanket or the like.
- This freedom of positioning permits a tighter control of the web printing feedback loop and thus guarantees a higher quality product than that possible with the prior art methodologies.
- the present invention need not be placed in these positions and may be placed elsewhere in the printing process as are some conventional inspection systems.
- the present invention is the only system/method available that permits the inspection loop to begin when the web ink is still wet, or when the web is experiencing velocity variations , flutter , and/or a change in focal distance between the web and the inspection sensor.
- FIG. 23 illustrates a commercial web offset press.
- This press configuration is identical to the newspaper press of FIG. 20 with the exception that it includes a dryer (2307, 2308) and chill rollers (2311, 2310, 2312).
- the dryer and chill rolls are for curing inks that are printed on coated stocks on both sides of the web and where there is little absorption as there is when printing on newsprint.
- the high speed printed web remains in the dryer for sufficient time to heat the web after which it passes over the chill rolls which sets to permanently hardens it.
- This curing process is referred to as "heat set" web offset printing and is used in all high quality printing on glossy or coated stocks as in the printing of magazines.
- the first location where a backup roller can be located is at the chill rolls where presently web scanners using photocells or linear arrays would be located.
- the position of these web scanners is shown as web scanner (2312) scanning the top side of the web, and web scanner (2310) scanning the bottom side of the web.
- FIG. 24 illustrates an expanded view of FIG. 23 with cameras (2305, 2306) of FIG. 23 represented by (2401, 2402) of FIG. 24, print unit (2304) of FIG. 23 represented by (2403) of FIG. 24 and a portion of dryer (2307, 2308) of
- FIG. 23 represented by (2404, 2405) of FIG. 24 respectively.
- the change in the field of view for a 0.10 inch change in focal distance (2508) is 0.006255-inch for a field of view (2507) equal to 0.5 inch, and for a focal distance (2506) equal to 4 inches.
- the cameras can be located to view marks at positions (2305, 2306) of FIG. 23 without backup rollers since these focal distance changes are automatically calibrated each time a mark pattern image is obtained using duplicate marks.
- the cameras can be located to scan unsupported web as in FIG. 24 with accuracy unaffected by web flutter or other focal distance variations such as just explained. In this instance with the cameras located this close to the last printing station additional waste can be eliminated with early detection and correction of errors instead of waiting until the web reaches the chill rolls as is current practice in photocell or linear array register controls.
- Plate Skew Plate skew errors are register errors due to non-parallelism of the printing plate across the entire web width. They are introduced in the manufacture of the plates or during the process of mounting new plates. Some presses have a motorized skew adjustment which allows skewing of the printing cylinder. However, with more accurate plate registration and mounting systems, plate skew errors are small enough that the operator can make adjustments to split the errors equally on both sides of the web.
- Fan-out is a phenomenon evidenced by minute lateral changes in the printed image that appear from the operator side to the gear side of the printed web. It is called fan-out because the lateral image width appears to increase after it is first printed. As the web proceeds from the first to the last printing unit, it undergoes a transition due to the addition of water, impression pressure, and tension that tend to relax it. This produces a slight lateral growth that becomes narrower at each succeeding printing station. It is presently corrected using either a wheel that indents the web tending to reduce the width of the printed image minutely, or by profiling a tension across the web before it enters the first printing station. While neither of these mechanisms is discussed in detail within this disclosure, they are widely known and used in the printing industry. Suffice it to say that the complete two dimensional register system in this disclosure would require motorized mechanisms for fan- out if a completely automatic two dimensional register control is to be implemented.
- FIG. 24 shows a two-camera system that for this application will be the motorized two-camera system of FIG. 5.
- the traverses (503, 504) are very rugged linear mechanisms of the same type that is typically used for very accurate positioning in machine tools.
- the complete mechanical assembly is designed for permanent mounting and is squared to the press so as to be able to measure skew of the printing cylinder and plate.
- FIG. 26 shows the mark patterns as first observed and before any register adjustment are made. It is assumed for this discussion that any overlapping marks will have been corrected as described elsewhere in this disclosure. Thus, all three errors will be present as shown in mark patterns (2601) on the operator side and mark pattern (2602) on the gear side of the press in FIG. 26. For purposes of simplification and clarity of explanation, the process will be broken down and described in three steps:
- FIG. 28 Both of these images are shown in FIG. 28 with the aligned image (2701) of FIG. 27 shown as marks (2805, 2821, 2822, 2823, 2824) of FIG. 28 and the image of mark pattern (2702) of FIG. 27 shown as marks (2804, 2817, 2814, 2815, 2816) of FIG. 28.
- the skew of the reference marks (2803) is determined by measuring the number of pixels in the X direction to the center of mark (2805) , moving the traverse to the other side of the web at the same strobe encoder setting and measuring the distance to the center of mark (2804) .
- the difference (2803) of these two values represents the skew of the printing plate and any skew or parallel misalignment of the traverse and printing cylinder.
- the skew of each mark is represented by the difference in X values or (2810-2806) as the skew of mark (2817) to mark (2821), (2811-2807) as the skew of mark C2818) to mark (2822) , and (2812-2808) as the skew of mark (2819) to mark (2823) .
- Some presses are equipped with motorized cylinder skew adjustments that provide a means for automatic skew adjustment with the skew values just calculated introduced directly by adjusting these motors. Most presses do not have this adjustment and thus the skew errors are adjusted equally between each side by moving the circumferential register mechanism.
- the Y dimensions (2814) of mark (2817), (2815) of mark (2818), and (2816) of mark (2819) represent fan-out of each color or the growth of the printing width through the press. Fan- out can be partially or totally corrected with several mechanical devices that are well known in the industry. Needless to say these devices must be motorized if a full two-dimensional initial automatic register alignment is to be performed.
- FIG. 29 illustrates an exemplary representation of a three mark pattern printed on both edges of the web and used for making a two-dimensional preregistration on a press equipped with motorized circumferential, lateral, and/or fan-out correction mechanisms.
- FIG. 79 is a corresponding exemplary process flowchart corresponding with the following detailed description.
- Duplicate marks with one additional mark are shown to describe the process. Any number of additional marks can be printed with the same process to be described applied to each additional mark.
- a camera takes an RGB picture of the Gear-side marks and determines the [X,Y] locations of each of the desired marks.
- a motor driven traverse with accurate encoder positioning is used to move the same camera and take an RGB picture of the Operator-side marks. The [X,Y] locations of these marks are determined.
- the pictures taken in each case use the same strobe firing point on the printed image that is synchronized by a cylinder encoder that corresponds one- to-one with the printed image.
- the center coordinates of each mark are measured in an absolute coordinate system based on a fixed origin in two- dimensional space lying in the plane of the printed web.
- Each [X,Y] coordinate pair is an absolute [X,Y] distance in inches from the absolute origin.
- the X-origin is defined to be the center of the camera image while the camera is positioned to one extreme of the traverse mechanism.
- the Y- origin is chosen to be the center of the camera image when the strobe is firing at a position that will allow marks on the Gear- and Operator-side to be fully detectable by image processing software.
- XG1 Absolute X coordinate of the center of the first Duplicate mark (2911)
- YG1 Absolute Y coordinate of the center of the first Duplicate mark (2912)
- XG2 Absolute X coordinate of the center of the test station mark (2931)
- YG2 Absolute Y coordinate of the center of the test station mark (2932)
- X02 Absolute X coordinate of the center of the test station mark (2941)
- Y02 Absolute Y coordinate of the center of the test station mark (2942)
- XGT2 Desired absolute X coordinate of the center of the test station mark (gear-side) (2951)
- YGT2 Desired absolute Y coordinate of the center of the test station mark (gear-side) (2952)
- X0T2 Desired absolute X coordinate of the center of the test station mark (operator-side) (2961)
- Y0T2 Desired absolute Y coordinate of the center of the test station mark (operator-side)
- Skew Error Determination (7906) The skew error is a Y offset error from Operator to Gear side that results from inaccurate plate mounting. To calculate the skew error, two constants must be defined:
- YTSG Y amount of skew error due to the traverse on the Gear-side
- YTSO Y amount of skew error due to the traverse on the Operator-side
- the skew error is calculated like this for the first duplicate mark:
- YS1 Y skew error from gear to operator sides in inches of first duplicate mark
- the fan-out error is the amount the web will stretch in the operator-to-gear direction as it passes through the press. This is an X error measurement. To calculate fan- out for a desired test mark, the following calculations must be made:
- XW1 X distance in inches from the first duplicate mark on the Operator-side to the first duplicate mark on the Gear-side
- XGT2M Modified X desired location for the center of the test mark
- YGT2M Modified Y desired location for the center of the test mark
- XE2M X distance error from current test mark position to desired modified position
- YE2M Y distance error from current test mark position to desired modified position
- Each mark pattern consists of one or more marks printed by each printing station. If more than one mark is printed it must conform to one of the following mark pattern types: 1. Single Column,
- Duplicate Marks must generally occupy the extreme locations of the pattern. In a single column or single row pattern, there are two Duplicate Marks. In a multi-column or multi-row pattern, there are four Duplicate Marks.
- FIG. 31 illustrates a flowchart of the overall process of Mark Recognition (3100) .
- the input to this system is an RGB image of the marks being printed (3101) .
- the first step in the overall recognition process is to find the Duplicate Marks.
- the system assumes high contrast for these marks and therefore starts with a very conservative threshold to discretize (binarize) the RGB image. This is accomplished by comparing each Red, Green, and Blue Trixel with the starting threshold (ST) to compute the binary image (3102).
- ST starting threshold
- the next step is to create a list of possible Duplicate Marks (3103) . This list will be called the GOODLIST and the process of creating this list is outlined in FIG. 32.
- FIG. 41 outlines the process for finding the Duplicate Marks from the GOODLIST. With the Duplicate Marks found, all non-Duplicate Marks must be detected (3105).
- FIG. 50 outlines this process. With as many marks as possible detected, the next step is to match printed stations from the mark pattern specification to the detected marks (3106). This process is outlined in FIG. 51. If this step (3106) passes, a final check is made between the marks (3107) to see if any erroneous print exists. This process is outlined in FIG. 54.
- the first step in GOODLIST creation is to calculate some important pixel dimensions (3201) that will be used to analyze the image.
- FIG. 33 the process of calculating pixel dimensions begins. Given the camera type and zoom magnification used when obtaining the RGB image, the X/Y Pixel/Inch parameters are obtained from a lookup table (3301) .
- the desired WIDTH and LENGTH of the register marks (in inches or millimeters) are read from the mark pattern specification (3302) and scaled to pixel dimensions (3303) .
- the amount the WIDTH and LENGTH can vary is also read from the installation constants (3304) and scaled to their pixel dimensions (3305) .
- XMINDIP and YMINDIP are also calculated at this point (3306) . These values provide a means for allowing small "holes” in marks to pass through the recognition process.
- the last two variables XBOUNDARY and YBOUNDARY are read from the installation constants (3307) and scaled to pixel dimensions (3308) . At this point FIG. 33 exits back to FIG. 32 (3309) .
- the process of GOODLIST creation continues by taking a series of evenly spaced vertical slices through the binary image and attempting to recognize marks of the desired size.
- the vertical mark size counter is also zeroed at this point (3203) . If the read pixel is a one (3204) , this means we are on a mark and therefore should increase our vertical mark size counter (3206) . If the pixel is a zero (3204) , this means we are on the background, and need to process a "vertical dip" (3207) .
- FIG. 34 illustrates the process of analyzing a "vertical dip” (3400) .
- a “dip” is a term used here to describe a potential mark hole. These holes are allowed through the process as still belonging to a potential mark as long as they do not exceed XMINDIP or YMINDIP (3306) .
- the mark size counter is tested to see if a mark was in progress (3401) . If a mark was not in progress, FIG. 34 is exited (3413) . If a mark was in progress, a counter called DIP COUNT is set to 1 (3402) to indicate that there is one pixel so far belonging to a "dip".
- the Y coordinate is checked to make sure there are still more lines (3403) . If there are no more lines, FIG. 34 is exited (3413) by first clearing the mark size counter (3412) . The next line pixel is read and checked to see if it lies on a mark (3407) . If it does, the "dip” has passed through and FIG. 34 is exited (3408) . If the next line pixel lies on the background (3407), the DIP COUNT is checked to see if it exceeds the threshold for vertical "dips", YMINDIP (3409) .
- the mark size counter is incremented (3404) , and the loop continues at line count check (3403) . If the DIP COUNT does exceed YMINDIP (3409) , it is certain now that the background has been detected and the preceding mark size count may be examined as a potential mark. The mark size count is checked (3410) to see if it falls inside the LENGTHVAR (3305) around LENGTH (3303) . If it does not, this vertical mark size does not belong to a mark of interest and FIG. 34 exits (3413) by first clearing the mark size counter (3412) . If the vertical mark size is correct, the next step is to process the horizontal direction to find the horizontal mark size (3411) .
- FIG. 35 illustrates the process of computing the horizontal mark size (3500) .
- the [X,Y] starting coordinates are computed (3501) .
- the X coordinate is the X value used in FIG. 32 for the vertical slice, and the Y coordinate is the center of the vertical slice (3501) .
- First the Right edge of the potential mark is found (3502) .
- FIG. 36 illustrates the process of finding the right edge of the potential mark (3600) .
- the first pixel is read from the binary image at coordinate [X,Y] (3601).
- a RIGHT COUNT counter is zeroed (3602) .
- the pixel is checked to see if it lies on a mark (3604) . If is does the RIGHT COUNT is incremented (3605) . If the pixel lies on the background, the a right horizontal "dip" is processed (3606) .
- FIG. 39 illustrates the right horizontal "dip” processing (3900) .
- First the RIGHT COUNT is checked to see if a mark was being processed (3901) . If not, FIG. 39 exits (3912) by first clearing the RIGHT COUNT (3911) .
- a DIP COUNT variable is set to one (3902) to indicate that a single pixel lies on a "dip". If the RIGHT COUNT does indicate that a mark was being processed, a check is made to see if any more right pixels exist in this slice on the image (3903) . If not, FIG. 39 exits (3912) by first clearing the RIGHT COUNT (3911) . If there are more right pixels (3903) , the next right pixel is read from the image (3905) .
- FIG. 39 exits (3907) . If the next right pixel lies on the background (3907) , the DIP COUNT is checked to see if it exceeds the threshold for horizontal "dips", XMINDIP (3908) . If it does not, the DIP COUNT is incremented (3909), the RIGHT COUNT is incremented (3904) , and the loop continues at right pixel check (3903) . If the DIP COUNT does exceed XMINDIP (3908), it is certain now that the background has been detected and the preceding RIGHT COUNT may be examined as a potential mark.
- the RIGHT COUNT is checked (3910) to make sure it is not larger than the WIDTHVAR (3304) plus WIDTH (3303) allow. If it is larger, this mark size does not belong to a mark of interest and FIG. 39 exits (3912) by first clearing the RIGHT COUNT (3911) . If the RIGHT COUNT is correct, FIG. 39 is exited (3912) and the RIGHT COUNT is not cleared.
- FIG. 36 continues by checking to see if there are no more right pixels to process (3607) . If there are no more, FIG. 36 exits (3608) with whatever RIGHT COUNT is left. If there are still more pixels, the next right pixel is fetched from the image (3603) and the loop continues at pixel mark check (3604) .
- FIG. 37 illustrates the process of finding the left edge of the potential mark (3700) .
- the first pixel is read from the binary image at coordinate [X,Y] (3701).
- a LEFT COUNT counter is zeroed (3702) .
- the pixel is checked to see if it lies on a mark (3704) . If is does the LEFT COUNT is incremented (3706) . If the pixel lies on the background, the a left horizontal "dip" is processed (3705) .
- FIG. 40 illustrates the left horizontal "dip” processing (4000) . First the LEFT COUNT is checked to see if a mark was being processed (4001). If not, FIG. 40 exits
- FIG. 40 exits (4012) by first clearing the LEFT COUNT and RIGHT COUNT (4011) .
- a DIP COUNT variable is set to one (4002) to indicate that a single pixel lies on a "dip". If the LEFT COUNT does indicate that a mark was being processed, a check is made to see if any more left pixels exist in this slice on the image (4003) . If not, FIG. 40 exits (4012) by first clearing the LEFT COUNT and RIGHT COUNT (4011) . If there are more left pixels (4003) , the next left pixel is read from the image (4004) . If this pixel is found to lie on a mark (4006) , FIG. 40 exits (4007) .
- the DIP COUNT is checked to see if it exceeds the threshold for horizontal "dips", XMINDIP (4009) . If it does not, the DIP COUNT is incremented (4008), the LEFT COUNT is incremented (4005), and the loop continues at left pixel check (4003) . If the DIP COUNT does exceed XMINDIP (4009), it is certain now that the background has been detected and the preceding LEFT COUNT may be examined as a potential mark. The LEFT COUNT plus RIGHT COUNT is checked (4010) to make sure it falls within the WIDTHVAR (3305) and WIDTH (3303) specification.
- FIG. 40 exits (4012) by first clearing the LEFT COUNT and RIGHT COUNT (4011) . If the size is correct, FIG. 40 is exited (4012) and the LEFT COUNT and RIGHT COUNT are not cleared.
- FIG. 37 continues by checking to see if there are no more left pixels to process (3707) . If there are no more, FIG. 37 exits (3708) with whatever LEFT COUNT and RIGHT COUNT are left. If there are still more pixels, the next left pixel is fetched from the image (3703) and the loop continues at pixel mark check (3704) . Mark Width Calculation
- FIG. 38 illustrates the process of performing a quick boundary check (3800).
- First the potential mark's bounding box is computed (3801) . This is the box which entirely encloses the potential mark: (xs,ys), (xe,ye).
- Next the boundary box is computed: (xl, yl) , (x2, y2) . (3802). This box is larger in all directions from the bounding box by
- FIG. 38 exits with this mark not being added to the GOODLIST (3811). If this check passes (3807), the Prgb[n] trixel is saved in the Srgb[n] trixel (3808).
- FIG. 38 exits (3811) by first adding this mark to the GOODLIST (3810) .
- FIG. 35 a mark has or has not passed the quick boundary check (3506) .
- FIG. 35 exits (3507) .
- the horizontal direction is fully processed (3411) and FIG. 34 exits (3413) by first clearing the MARK COUNT (3412).
- a vertical "dip" has been fully processed (3207) , and a check is then made to see if any more lines exist in this vertical slice (3208) . If so, the next line pixel is read from the binary image (3205) and the vertical slice loop continues at the mark check (3204) .
- FIG. 31 illustrates this process (4100) .
- FIG. 42 illustrates the calculation of these constants (4200) .
- Millimeters is read (4201) . This is an installation constant and forces the duplicate marks to be either vertically or horizontally collinear by this amount. This value is scaled into the pixel dimensions: XCOLLINEARITY and YCOLLINEARITY (4202) . An installation constant, COLORVAR, is read next (4203) . This variance is a measure of how different the color of one Duplicate Mark to the other may differ before they are not considered Duplicate Marks.
- the values DUPX and DUPY are read from the mark specification (4204) . These are the absolute distances, center-to-center, desired between the Duplicate Marks. These values are scaled to their pixel dimensions (4205) .
- the WINDOWWIDTH and WINDOWLENGTH constants are also obtained from the mark specification (4206) . These are the overall width and length (Inches or Millimeters) of the mark specification. These values are scaled to their pixel values (4207) .
- Another set of installation constants, XDUPBOUNDARY and YDUPBOUNDARY, are read (4208) . These constants define the X and Y area thickness around the outside perimeter of the Duplicate Marks that must be free of erroneous print to qualify as Duplicate Marks. These values are then scaled to pixel values (4209).
- FIG. 42 exits back to FIG. 41 (4210).
- the comparison constants have now been computed (4101) .
- the next step is to determine the type of mark pattern being printed in order to process the Duplicate Marks correctly. If the mark pattern is the Single Column Mark Pattern (4102), FIG. 43 is employed. If the mark pattern is the Single Row Pattern (4103) , FIG. 44 is employed. If the mark pattern is the Rows > Columns Pattern (4104), FIG. 45 is employed. If the mark pattern is the Columns > Rows Pattern (4105) , FIG. 46 is employed. If the mark pattern is a Single Mark Pattern
- FIG. 47 is employed. If Duplicate Marks are detected (4107) , FIG. 41 exits back indicating success (4108) . If the marks were not detected, FIG. 41 exits back indicating failure (4109) .
- FIG. 43 illustrates the process of detecting Duplicate Marks that belong to the Single Column Mark Pattern (4300) .
- marks on the GOODLIST are checked against a set of rigorous collinear, dimensional, and color checks (4303).
- FIG. 48 illustrates this process. Refer below to "Finding the Collinear Marks", for a description of this process. If the Duplicate Marks are found (4304), FIG. 43 exits with success (4305). If not, FIG. 43 exits with failure (4306) .
- FIG. 44 illustrates the process of detecting Duplicate Marks that belong to the Single Row Mark Pattern (4400) .
- marks on the GOODLIST are checked against a set of rigorous collinear, dimensional, and color checks (4403).
- FIG. 48 illustrates this process. Refer below to "Finding the Collinear Marks", for a description of this process. If the Duplicate Marks are found (4404), FIG. 44 exits with success (4405). If not, FIG. 44 exits with failure (4406) .
- FIG. 45 illustrates the process of detecting Duplicate Marks that belong to the Rows > Columns Mark Pattern (4500) .
- marks on the GOODLIST are checked against a set of rigorous collinear, dimensional, and color checks (4503).
- FIG. 48 illustrates this process. Refer below to "Finding the Collinear Marks", for a description of this process. Use the first Duplicate Mark found (DUP1) as a starting point and set the direction to HORIZONTAL (4504) .
- FIG. 45 exits with success (4510) . If not, FIG. 45 exits with failure (4509) .
- FIG. 46 illustrates the process of detecting Duplicate Marks that belong to the Columns > Rows Mark Pattern (4600) .
- marks on the GOODLIST are checked against a set of rigorous collinear, dimensional, and color checks
- FIG. 48 illustrates this process. Refer below to
- FIG. 47 illustrates the process of -detecting the single mark from the GOODLIST (4700) .
- Get the first box on the GOODLIST call it TESTBOX (4704) . If this is not the same box as BOX (4705), calculate the center of TESTBOX (txcen, tycen) (4706).
- FIG. 48 illustrates this process of finding collinear marks (4800) .
- First the first box on the GOODLIST is fetched and labeled BOX (4802) .
- FIG. 48 exits with success (4814) .
- This process is illustrated in FIG. 49 and below in "The Duplicate Mark Boundary Check" section. If the DUP Boundary Check fails (4813) , the previous tests failed (4809) , or BOX was the same box as TESTBOX (4807), get the next TESTBOX on the list (4810) . If there are more boxes for TESTBOX (4811) , continue looping at the same box test (4807) .
- FIG. 49 illustrates the Duplicate Boundary check process (4900) . First the initial boundary offsets are set
- FIG. 49 exits with success
- FIG. 50 illustrates this process (5000) .
- First the region of interest is calculated (5001) . This is the mark pattern specification window circumscribed around the duplicate marks.
- Set the threshold to the starting threshold (ST), an installation constant (5002) .
- If the ending threshold has been reached (5007) , an installation constant, FIG. 50 exits with failure (5008) .
- the threshold has not reaching the ending value (5007) , continue looping at the create binary image step (5003) . As soon as enough marks are detected (5005), FIG. 50 exits with success (5009) .
- FIG. 51 illustrates this process (5100) .
- First the region of interest is calculated (5101) . This is the mark pattern specification window circumscribed around the duplicate marks. If the mark pattern is either the Single Row type or the Columns > Rows type (5102) , match the marks to stations for these types (5103) .
- FIG. 52 illustrates this process (5200) .
- the region of interest is divided into rows of equal size (5201) .
- the first row is selected as the region of interest (5202) . If more marks are needed for stations in this row (5203) and more marks exist in this row (5205) , match a mark to a station from left to right in the row
- FIG. 52 exits with success (5208).
- FIG. 52 exits with failure (5209).
- an attempt at matching a Single Row or Columns > Rows pattern was attempted (5103) . If this was a good match (5105) , the marks for each station are sampled inside (5108) and outside (5109) to determine the interior and exterior color. FIG. 51 then exits with success (5107). If this was a bad match, FIG. 51 exits with a failure (5106) .
- FIG. 53 illustrates this process (5300) .
- First the region of interest is divided into columns of equal size (5301) .
- the first column is selected as the region of interest (5302) . If more marks are needed for stations in this column (5303) and more marks exist in this column (5305) , match a mark to a station from top to bottom in the column (5306) and continue looping at the more marks are needed step (5303) . If more marks are not needed for stations in this row (5303) and more marks do not exist in this column (5304) , see if there are more columns (5307) . If there are more columns (5307) , continue looping at the set next column as region of interest step (5302) . If there are no more columns to process (5307), FIG. 53 exits with success (5308).
- FIG. 53 exits with failure
- FIG. 51 Referring back to FIG. 51, an attempt at matching a Single Column or Rows > Columns pattern was attempted (5104) . If this was a good match (5105) , the marks for each station are sampled inside (5108) and outside (5109) to determine the interior and exterior color. FIG. 51 then exits with success (5107) . If this was a bad match, FIG. 51 exits with a failure (5106) .
- FIG. 54 illustrates this process (5400) .
- FIG. 55 illustrates this process (5500) .
- First the region of interest is divided into rows of equal size (5501) .
- the first row is selected as the region of interest (5502). Get the first station's mark that was matched in this row (5503) . If there are more stations to process in this row (5504) , get the next station mark that was matched in this row (5505) . Check for erroneous print in the area between these marks (5507) . If there is some print (5508) , FIG. 55 exits with failure (5510) . If there is no print (5508) , save this mark location (5509) and continue looping at the check form more stations in this row step (5504) . If there are no more stations in this row (5504) , see if there are more rows to process (5506) .
- FIG. 55 exits with success (5511) . If there are more rows (5506) , continue looping at the set row region of interest step (5502) . Referring back to FIG. 31, an attempt at checking between the marks for a Single Row or Columns > Rows pattern was attempted (5403). If this was a good check (5405), FIG. 54 exits with success (5406) . If this was a bad check, FIG. 54 exits with a failure (5407) . If the pattern (5402) was of type Single Column or Rows > Columns, try to match this type of pattern (5404) . FIG. 56 illustrates this process (5600) .
- the region of interest is divided into columns of equal size (5601) .
- the first column is selected as the region of interest (5602). Get the first station's mark that was matched in this column (5603) . If there are more stations to process in this column (5604) , get the next station mark that was matched in this column (5605) . Check for erroneous print in the area between these marks (5607) . If there is some print (5608) , FIG. 56 exits with failure (5610) . If there is no print (5608) , save this mark location (5609) and continue looping at the check form more stations in this column step (5604) . If there are no more stations in this column (5604) , see if there are more columns to process (5606). If there are not, FIG. 56, exits with success (5611) . If there are more columns (5606) , continue looping at the set column region of interest step (5602) .
- FIG. 54 exits with success (5406). If this was a bad check, FIG. 54 exits with a failure (5407) .
- the purpose of the light mark detection software is to find faint registration marks given an RGB image including a rectangular region of interest (ROI) including the faint marks and two duplication marks.
- the process is multi-step:
- FIG. 60 illustrates an exemplary system flowchart of the overall process of finding light marks within a given ROI.
- the detection process begins when the control program passes a Red-Green-Blue (RGB) image to the detection program
- RGB Red-Green-Blue
- Each color plane (Red, Green, and Blue) is a gray level image digitized (typically to 8 bits resolution)
- the input Red-Green-Blue (RGB) images are first used to calculate corresponding Hue-Saturation-Intensity (HSI) images (6003). These calculations are performed on a pixel- by-pixel basis. The formulas for HSI calculations are detailed in FIG. 67. These six separate images are used as input to the segmentation process (6004) .
- 61 is used to extract the d-marks from the rest of the ROI.
- LMI Light Mark
- the next step is to image process the LMI for light marks (6005) .
- This process is detailed in FIG. 62.
- the result of this process is to create the Result Image (RI) that can be analyzed for light mark measurement.
- the final step is to label and analyze the image processed map for light marks (6006) .
- FIG. 61 An exemplary light mark segmentation process is detailed in FIG. 61.
- the basic idea behind segmenting out the d-marks from the rest of the image is variation in intensity.
- the intensity of an image is a measure of the image darkness or lightness. Since the d-marks are the darkest objects in the ROI passed to the segmentation software, they can be segmented by finding the lowest intensity objects in the ROI.
- the maximum of the RGB planes is determined (6102) . This is done by comparing the Red, Green, and Blue planes on a pixel-by-pixel basis. Thus, for a particular pixel position in each of the RGB planes: Pr, Pg, Pb, the maximum valued pixel is found and written to a Max Plane (MP) . That is if Pr > Pg and Pr > Pb, Pr is written to MP. If instead Pg > Pr and Pg > Pb, Pg is written to MP. Finally, if Pb > Pr and Pb > Pg, Pb is written to MP. This is done for every corresponding pixel in the RGB planes.
- the resulting MP plane is now gray level eroded (6103).
- This process is detailed in FIG. 68.
- the process begins at (6800) when the subroutine is called. Any input image can be eroded.
- the MP plane is the input plane to be eroded and is set equal to the IM plane (6801). Every pixel in the IM plane is considered in gray erosion.
- the first pixel P[n] is first acquired (6802) and then the eight neighboring pixels P[n-4], P[n-3], P[n-2], P[n-1], P[n+1], P[n+2], P[n+3], P[n+4] (6803).
- the minimum pixel value of these 9 pixel values is determined (6804), that is the smallest pixel value in this group of 9 is determined and then replaces pixel P[n] (6805) . If there is another pixel to be analyzed (6806), the next pixel in IM is then acquired (6807), set to P[n] and program control goes back to (6803) where the 8 neighbor pixels are determined. This process continues until all pixels in IM (MP) have been analyzed.
- the next step is to gray dilate the MP plane (6104) .
- This process is detailed in FIG. 69.
- the process begins at (6900) when the subroutine is called. Any input image can be dilated.
- the MP plane is the input plane to be dilated and is set equal to the IM plane (6901). Every pixel in the IM plane is considered in gray dilation.
- the first pixel P[n] is first acquired (6902) and then the eight neighboring pixels P[n-4], P[n-3], P[n-2], P[n-1], P[n+1], P[n+2], P[n+3], P[n+4] (6903).
- the maximum pixel value of these 9 pixel values is determined (6904), that is the largest pixel value in this group of 9 is determined and then replaces pixel P[n] (6905). If there is another pixel to be analyzed (6906), the next pixel in IM is then acquired (6907), set to P[n] and program control goes back to (6903) where the 8 neighbor pixels are determined. This process continues until all pixels in IM (MP) have been analyzed.
- Convolution requires an input image (IM) on which to perform the convolution, an input 3x3 convolution kernel (K) that specifies the type of filtering to perform on IM and the convolution shift value (S) that is the divisor for each sum of product operation in the convolution. S normalizes the resulting or convolved image .
- IM input image
- K input 3x3 convolution kernel
- S convolution shift value
- our input image is the MP plane, which is set equal to IM for this operation (7001) , K is set to a 3x3 convolution kernel of all 1 ' s which is a low pass filtering operation, and S is set to 9 which results in a convolved image with the same intensity range as the input image.
- IM the input image
- the next step is to get the eight neighbors of P[n]: P[n-4],
- Ptot is the sum of the nine image pixels P[n- 4]...P[n+4] since the kernel values are all 1.
- the next step is to replace the original input pixel P[n] with the normalized sum of products (7005) .
- the next step is to check to see if there is another pixel that can be processed
- Histogramming is a method of determining the pixel intensity value distribution in an image.
- First the input image MP is set to the subroutine input image IM and an array of 256 integer values, 1 value for each possible intensity value in an image is created (7101) .
- Next the first pixel (P[n]) in the input image (IM) is obtained (7102).
- the value of P[n] serves as the address into the Histogram Array (HA) .
- the integer value stored at this address is incremented by 1 (7103) .
- Next the image is checked to see if there is another pixel that can be processed (7104) . If there is, the next pixel value P[n] is obtained (7105) and this new pixel value serves as the address into the HA as before (7103) . If there are no more pixels to process in IM, the routine is exited with the HA returned.
- the next step is to determine the threshold value for the lowest intensity in the image and threshold the MP plane using this value (6107). This is accomplished as follows: the HA array is analyzed to find the first maximum or "bump". This maximum is assumed to be the d-marks in the image. The half power point of this maximum is then determined by finding the right-hand valley corresponding to this maximum, this is the d-mark threshold value (Dthresh) . Then the MP plane is thresholded such that all pixel values in MP that are less than Dthresh are set to 0 and all pixel values greater than or equal to Dthresh are set to 1 (6108) . The result of this operation is the d-mark Map (DMM) that has a pixel value of 0 where the d-marks are in the ROI and 1 everywhere else in the ROI where the ROI should be analyzed for light marks.
- DDMM d-mark Map
- the next step is to logically AND the input intensity (I) image with the DMM (6109) to create the Light Marks Image (LMI).
- I input intensity
- LMI Light Marks Image
- the image processing of the LMI begins at (6200) with the input LMI (6201) from the previous segmentation process.
- the input LMI image is now enhanced (6202). This is accomplished as shown in FIG. 72.
- the idea is to find the actual difference between the lightest and darkest pixels in the I-image and "stretch" this range through the entire range of possible values which is 256 because each pixel is digitized to 8 bits.
- the subroutine (7200) is entered and sent the input image, the LMI plane in this case equal to the Image Input Plane (IM) (7201) .
- the next step is to histogram IM (7202) in the same manner as already discussed in FIG. 71.
- the next step is to gray erode the enhanced LMI plane (6203).
- This process has already been detailed in FIG. 68.
- the effect of the process is to group darker valued pixels together thus making a darker mark area more homogeneous.
- the next step is to low pass filter the enhanced LMI plane (6204) .
- This process has already been detailed in FIG. 70.
- a 3x3 kernel of all l's and a shift value of 9 is typically used. This operation results in all pixels in the image being replaced by the average of those pixels and their 8 neighbors thus "blurring" or low-passing the images. This process has the effect of making the lighter and darker areas in the image even more homogeneous.
- the result of this operation is the results image (RI) .
- the next step is to histogram the RI plane after these operations (6205) . This process has already been discussed in FIG. 71.
- the input image is an enhanced image of light substrate with various (potentially) darker marks
- a histogram with a minimum of 1 major lobe will be obtained.
- the lobe closest to the maximum (256) will correspond to the substrate because the substrate will have a higher intensity value than the darker marks.
- the next step is to determine the substrate threshold value (Sthresh) that is the peak value position of the upper lobe (6206) .
- the next step is to obtain the dark threshold value (Dthresh) by finding the left hand valley of the upper lobe (the point where the number of set pixels is negligible in the histogram) and dividing this value by 2 (6207) .
- This value represents a safe value to begin thresholding the RI map and insure the light marks have not been thresholded out.
- the final step (6208) is to return to the calling routine with both upper and lower thresholds (Sthresh & Dthresh) and the RI image.
- the object is to now search for light marks in the RI plane (6006) .
- This process is detailed in Figs. 63 - 66.
- the light mark detector is started (6300), inputs are the RI plane, Sthresh, Dthresh, nMarks, minWidth, maxWidth, minHeight, maxHeight (6301).
- the LMM is an image consisting of only two pixel values: 0 for light marks and 1 for substrate.
- LMM LMM
- LCM label list
- the next step is to sort the LCM largest to smallest (6305) using an Imaging Tech supplied routine.
- the next step is to cycle through the LCM, getting the first (and largest) label. By reading the numeric label value for this label, it is possible to cycle through the LMM to find all pixels that belong to this label.
- the next step is to find the minimum X and Y and the maximum X and Y extents of the pixels of this label in the image using an Imaging Tech supplied routine. This gives a bounding box (CBB) which completely encompasses the label (6306) as well as the height, width, and center coordinates of the CBB: CBBHeight, CBBWidth, CBBCenter.
- CBB bounding box
- the next step is to determine if CBBHeight and CBBWidth fit within the mark min and max sizes input to the program: minWidth, maxWidth, minHeight, and maxHeight (6307) . If they do not (6309) , the LCM is checked for another label as detailed in FIG. 64 (6401), (6405), another label in the LCM is acquired (6406), (6402). Referring back to FIG. 63 (6310), this new label is checked as just detailed until the LCM is exhausted. If CBBHeight and CBBWidth fit within the mark min and max sizes input to the program (6308) another path is taken. Referring to FIG.
- CBBCenter is checked against a list of previously found marks: the Stored Mark Table (SMT) (6404). It is determined whether CBBCenter lies within any of the CBB's stored in the SMT. If it does exist in the SMT (6405) , the next label in the LCM is acquired and checked in the same manner as before. If CBBCenter does not lie within any of the CBB's stored in the SMT, a new light mark has been found. This CBB is added to the SMT (6407) . At this point, the SMT is checked to see if there are nMarks CBB's stored in it (6408). This check is also performed when the LCM has been exhausted of labels to check (6405) .
- SMT Stored Mark Table
- the idea of the process is to threshold and label the image containing the light marks at successively higher threshold values until all marks are found at their own, optimum threshold value. This insures that light marks will be found before lighter substrate imperfections.
- FIG. 65 which occurs at the conclusion of either nMarks CBB' s being in the SMT or Thresh incremented up to the substrate threshold Sthresh, the entire process is repeated but this time beginning the thresholding process at Sthresh and decrementing Thresh until nMarks CBB' s are again found in the SMT or Thresh is decremented down to the dark threshold Dthresh.
- the next step is to sort the LCM largest to smallest (6504) using an Imaging Tech supplied routine.
- the next step is to cycle through the LCM, getting the first (and largest) label.
- the next step is to find the minimum X and Y and the maximum X and Y extents of the pixels of this label in the image using an Imaging Tech supplied routine. This gives a bounding box (CBB) which completely encompasses the label (6505) as well as the height, width, and center coordinates of the CBB: CBBHeight, CBBWidth, CBBCenter.
- CBB bounding box
- the next step is to determine if CBBHeight and CBBWidth fit within the mark min and max sizes input to the program: minWidth, maxWidth, minHeight, and maxHeight (6506) . If they do not (6508) , the LCM is checked for another label as detailed in FIG. 66 (6601), (6605), another label in the LCM is acquired (6606), (6602). Referring back to FIG. 65
- CBBCenter is checked against a list of previously found marks: the Stored Mark Table (SMT) (6604). It is determined whether CBBCenter lies within any of the CBB's stored in the SMT. If it does exist in the SMT (6605), the next label in the LCM is acquired and checked in the same manner as before. If CBBCenter does not lie within any of the CBB's stored in the SMT, a new light mark has been found. This CBB is added to the SMT (6607) .
- SMT Stored Mark Table
- the SMT is checked to see if there are nMarks CBB's stored in it (6608) . This check is also performed when the LCM has been exhausted of labels to check (6605) . If there are nMarks in the SMT (6611) the SMT is returned to the calling program. If there are not nMarks in the SMT, the original threshold variable Thresh is decremented by 1 (6609), then checked to insure it is greater than Dthresh, the dark threshold (6610). If Thresh is less than Dthresh, the SMT is returned to the calling program. Otherwise, processing continues as described before in FIG. 65 (6510).
- An advantage of a web offset press is that copy changes can be made quickly and very inexpensively more so than with any other printing process.
- Each time a copy change is made new plates with the new copy must be changed.
- On most all presses there are errors in location of the plates that are caused by errors in plate making, errors in mounting the plates on the press and errors due to the mechanical mechanisms which correct register not being in their center position. These three sources of errors will show up immediately when the press starts printing. Normally for most all presses, the maximum error for all colors both in the lateral and circumferential directions are well within plus or minus 0.25 inch.
- a camera with a field of view of 0.5 inch in both the X and Y direction provides a 0.001-inch resolution for each pixel and is ideally suited for web offset automatic register applications.
- all marks will fall within the field of view of the camera.
- the initial register errors due to the above, can be significantly larger than the pattern with many of the marks falling outside of the duplicate marks as shown in FIG. 13.
- the operator would manually move all of the motors associated with the lateral and circumferential register mechanisms to bring all colors in alignment. Considerable waste material and loss press time is associated with this manual process costing many thousands of dollars per year.
- the following is a disclosure of a method for rapidly measuring the circumferential and lateral errors for each mark and an automatic method for moving the respective register motors so that all stations are within tolerance for the automatic final register system to operate.
- FIG. 13 A typical pattern of the mark positions upon mounting new plates for the mark pattern shown in FIG. 8 is shown in FIG. 13.
- Each mark usually is a different color and easily identified from the frozen image of the marks as shown in FIG. 13.
- Note 1 The marks are shown hatched in FIG. 13 to distinguish each mark as color images are not available for this disclosure.
- Note 2 The operator knows which color is associated with each printing tower.
- Each mark in the frozen image is selected by touching the center of the mark. Each time a display is presented which enables the operator to input the tower number for that mark. This procedure is performed for all marks within the pattern and in the frozen image. Based on this information the magnitude of correction is calculated to bring all stations into alignment or so that none of the marks overlap and all marks can be recognized by the software.
- the mark pattern is then presented on the display with each station selected for each mark position.
- a superimposed image of the mark pattern with each position of the mark pattern shown as a square box is superimposed over the frozen image and centered within the field of view.
- Each mark is dragged in turn to its respective box with the printing tower identified as in procedure 1 above.
- the purpose of the software is to characterize an input printed or cut object in multiple terms for later recognition.
- the printed or cut object could be such objects as a square mark or a straight cut.
- the object to be recognized is a square black mark on a white substrate.
- the Intensity (I) image of this object will be sufficient to adequately characterize this object.
- Input to the algorithm therefore, will be an I-image with a Region of Interest (ROI) encompassing the mark and surrounding substrate.
- ROI Region of Interest
- the algorithm makes conventional measurements on the object such as width, height, and location.
- a shape number is calculated for the object.
- the benefits of using shape numbers are many, but for our purposes, the great advantage is that shape numbers are size invariant.
- the shape number of a small rectangle is the same as that of a large rectangle. This property can be a great help in overcoming the effects of web jump - when the object is not fully represented in the image because the sync of the press has changed relative to the encoder sync on the camera system.
- FIG. 73 illustrates a flowchart of the overall process shape characterization.
- the process begins (7301) with the Intensity (I) image of the desired object to be characterized or recognized (7302) .
- the I-image is first segmented (7303) to separate the object from the substrate.
- This process is detailed in FIG. 74.
- the result of this operation is an Image Map where all pixels that correspond to the object are set equal to 1, and all pixels that correspond to the substrate are set equal to 0.
- the resultant image map is now labeled using Imaging Technology supplied routines (7304) .
- the result of this operation provides the location and dimension of a bounding box that completely encloses the object.
- standard measurements of the object are made using Imaging Technology supplied routines that operate on the bounding box (7305) .
- the standard measurements include width, height, centroid, perimeter, and area.
- the object is converted to a digitized object (7306). This process is detailed in FIG. 75. The result of this operation is much simpler version of the original object.
- the shape number of this digitized object is determined (7307) . This process is detailed in FIG. 76. Finally, all standard measurements and the shape number are stored and compared to previously stored members in a list. If a match is found in terms of shape numbers, further processing can be done using the standard measurements to determine the degree of probability that the object has been recognized before.
- the process begins (7401) with the input I-image containing the object and substrate
- the I-image is histogramed (7403) .
- This procedure has already been described in FIG. 71.
- the histogram in this case will be bi-modal with the lower maxima corresponding to the mark and the higher maxima corresponding to the substrate.
- the next step is to determine a mark threshold (7404) . This is determined by setting the threshold value, Othresh equal to the right hand half-power point of the maxima closest to zero.
- the I-map is thresholded at this threshold of Othresh (7405) .
- the result of this procedure is to provide an object map where pixels corresponding to the object are set to 1 and pixels corresponding to the substrate are set to 0.
- the object is digitized. This is to provide a simpler object to be shape classified in the next operation.
- the process is started (7501) with the image map and bounding box of the previous operation (7502) .
- a grid of given width and height is superimposed over the object (7503).
- Each position within the grid is analyzed.
- the first position (0,0) is obtained (7504), all object pixels within the grid (0,0) superimposed on the object map are counted.
- the ratio of counted object pixels to the total number of pixels is the object density within this grid position (7505).
- the corresponding grid position (in this case (0,0)) is set to 1 (7508) . Otherwise the grid position is set to 0 (7506) . The next grid position is obtained and the process is repeated for every grid position (7507, 7509) . At the conclusion of this operation, the digitized object (DO) has been created. This object is a much simpler representation of the original object in a lower resolution grid space.
- the digitized grid object of the previous operation is input to the shape number determination algorithm (7602) .
- this algorithm finds the perimeter of the digitized object and follows the perimeter until it reaches its starting point. As it traverses the object perimeter, it records each time it must make a left or right turn to stay on the object perimeter.
- a digitized object shape can be characterized as a sequence of left and right turns. For example, a square object would consist of four consecutive right hand turns and the shape number would therefore be 1111. As can be seen, shape numbers are rotational and size invariant.
- the object perimeter must be found. This is accomplished by starting at DO grid position (0,0) and incrementing each axis count by 1 until a non-zero value (1) is found at that grid location. This location is remembered as serves as the trigger to terminate the algorithm when the object perimeter has been completely traversed (7603) .
- the first four connected pixel starting from the arbitrarily chosen direction of preference (left) is found (7604) .
- Four connected pixels are pixels that are adjacent to a center pixel but only on cardinal points (north, south, east, and west). Diagonally adjacent pixels are not considered. If the change of direction from the first pixel to the first four connected pixel is to the left, the first shape number digit is set to 2 (7605) . If the first direction change is to the right, the first shape number digit is set to 1. If there is no change in direction, no shape digit is added.
- the next four connected pixel in DO starting from the arbitrarily chosen direction of preference (left) is found (7607) . This pixel position is checked to see if it is the original starting position, if so, the process is finished. If it is a new perimeter pixel, the shape digit for this direction change (if any) is determined and concatenated with the shape number.
- the purpose of the color monitoring software is to analyze the RGB color density of each mark and substrate (which is stored at mark recognition time) to see if a tolerance has been exceeded. If it has, an error flag is set and then passed back to the calling routine to warn the operator that a color density out-of-tolerance situation exists.
- FIG. 77 illustrates an exemplary system flowchart of the overall process of checking color density of the marks in the mark list (ML) found at registration time (7700) .
- the process begins by getting the first mark entry in the ML (7702) .
- all needed values (7703) are extracted for this mark entry:
- the current mark density is checked against the original mark density (7704) . If any of the following cases exist: abs (Mr-Mor) >Dvar, abs (Mg - Mog)>Dvar, abs (Mb- Mob)>Dvar, then an error flag is set (7706) and control is returned to the calling routine (7709) . Otherwise, the current substrate density is checked against the original substrate density (7705) . If any of the following cases exist: abs (Sr-Sor) >Dvar, abs (Sg-Sog) >Dvar, abs (Sb-Sob) >Dvar, then an error flag is set (7706) and control is returned to the calling routine (7709) .
- the ML is checked to see if another mark is in the list (7707) . If there is another mark in the ML, the next mark is retrieved (7708) and control returns to (7703) where the mark specifics are obtained and the process just described is repeated. If the
- a camera (304) is located on the delivery of the sheeter. This camera is magnified in FIG. 3 as (305) . The camera is shown viewing the edges of cut sheets on the delivery of the sheeter. Through image processing techniques the edges of the sheets are determined and the distance from the edges to a mark located near the edge is calculated from images that are continuously taken during production. If the distance between the edge of the cut sheet and the mark changes, correction signals are sent to the motor that moves the compensating roller (307) in the direction to reestablish this distance.
- FIG. 3 shows two additional cameras (304, 308) .
- Camera (308) is shown as viewing the edge of a cut sheet relative to a mark also shown as (1802) in FIG. 18.
- the other two images (1803, 1804) shown in FIG. 18 are the mark patterns printed on the top and bottom of the web and viewed by the two cameras contained in assembly (301) .
- obtaining and maintaining register and obtaining and maintaining color are the two major quality issues which also contribute to nearly all of the printing waste.
- the teachings of the present invention provide the means to significantly improve and maintain registration during initial setup and during the print run with a corresponding substantial reduction in overall printing waste.
- the same hardware described and used for registration and disclosed herein may also be utilized to both measure and control color both at start up and during the run. Since the traversing mechanism associated with a single camera is less costly than adding two more cameras for color measurement and control, this presents a very economical method of implementing this functionality within the context of existing web printing operations. Given that the hardware is common for both applications, the remaining components comprise additional software in addition with some interface logic to control the printing press color inking mechanisms.
- the ink keys mechanical controls in the printing press that determine the quantity of printing ink that is dispensed
- the ink keys can be set automatically for each set of new printing plates with a substantial reduction in setup time and material waste. This automatic control both eliminates the need for manual human intervention into the printing process (thus saving labor) , but also produces a substantially higher quality print product while simultaneously eliminating vast amounts of pre-registration printing waste.
- the web offset printing system is by far the most economic and highest quality process for printing process color reproductions.
- a minimum of four printing units are required and generally print the four-process colors yellow, cyan, magenta, and black. While the pigments used in the inks that print these colors are not pure, they are stable with sufficient quality control that they can reproduce over and over again the same results in color fidelity on the same stock. This is true for magazine printing as the same inks are always deposited on the same coated stock with the same degree of whiteness.
- the inks are used as delivered with no additions.
- the solid printed color is determined almost exclusively by the thickness of the ink film.
- This ink film is changed by adjusting an ink key that meters the ink to an area across the width of the printing press.
- an ink key covers a width of about one inch and thus a 38-inch wide press used extensively for printing magazines would have 38 ink keys per color unit.
- the adjustment of these ink keys is done either manually or by presetting the keys based on information obtained during the making of the plates.
- One way is to measure the inked areas on a plate when it is manufactured and then preset the ink keys based on these measurements .
- Densitometer The most common method used by every magazine printer to adjust the ink keys is based on densitometer readings taken from printed marks of the four colors. The ink keys are adjusted until a specific densitometer reading is achieved. This reading provides the proper ink film for the solid printed mark which then assured that the correct ink is metered to print the correct color in the area covered by that ink key. As long as the same inks and substrates are used, the densitometer reading of a solid printed mark will be the same regardless of the total ink used to print the area covered by that ink key. It is required to periodically recalibrate the densitometer using a white (uniform density) calibration plate supplied for that purpose.
- the densitometer readings for the solid printed marks will be the same. If the substrate is changed and has a different degree of whiteness, the densitometer reading will be different since the calibration white plate has not changed.
- FIG. 77 as described in the supporting text above provides a means to obtain RGB readings of the color and white background of any area, object, or mark. It would be desirable to convert the RGB readings from a RGB camera directly to densitometer readings thus easily replacing the densitometer and the manual means for setting the ink keys.
- An easier and more accurate means of providing densitometer readings is to provide a calibration plate just as was done in Zoom Lens Calibration to calibrate the Zoom Lens.
- the holders for the calibration plates are mounted on the traverse shown as (509) and (510) of FIG. 5. Item (509) is for the top camera (501) and item (510) is for the bottom camera (502) .
- a color calibration plate is installed on each calibration holder and read by each respective camera.
- the calibration plate includes as a minimum, a small patch of colors (yellow, magenta, cyan, and black) with a number of white patches of various degrees of whiteness. All of the patches are read using a precision densitometer with the densitometer reading stored in a lookup table in computer memory. Each of these same patches are then read by the camera and also stored in computer memory as RGB values. For each specific stock and ink, the densitometer readings and RGB readings are stored in a lookup table. This table is used to correct solid ink density measurements on the printing press corresponding to the RGB readings of the same marks taken by the camera from the samples read by the densitometer. A test would be run that purposely offsets the ink densities by manually adjusting the ink keys first in the lighter density direction and then in the darker density direction with both density and RBG readings stored in the computer lookup table.
- the camera could first read the whiteness of the substrate, and recalibrate itself from the background substrate or one of the white chips mounted on the calibration plate. This feature would provide the same density readings for all substrates.
- the procedure to adjust the ink keys automatically could utilize either the fixed lens system of FIG. 5 and/or the large field of view Zoom Lens System of FIG. 2 and/or any of the teachings of Zoom Lens Calibration.
- a set of marks like the register marks of FIG. 7 but separated by a much smaller distance would be printed one set for each ink key.
- the duplicate marks would provide the means of relocation of the mark sets for each ink key with rough location as defined by the encoder attached to the traverse.
- the color measurement system illustrated in FIG. 77 can be used as is without the need for a color calibration plate in this example. In these circumstances, a difference in color reading is to be monitored instead of an absolute color reading as in the initial adjustment of ink keys.
- the color of any number of areas or objects can be monitored with the color first adjusted to its correct value and then compared during printing for color variation.
- An offshoot of this technique to save time and material for obtaining the correct color of multiple patterns is to first adjust the baseline pattern either by eye or using a densitometer, and then having the automated control system described herein to duplicate the color on the remaining patterns by adjusting the ink keys automatically for the other patterns of the same color.
- the color of the register marks may be monitored according to the teachings illustrated in FIG. 77 and stored when the print job is first run.
- the color difference will be detected and a global change can be made to all affected ink keys. That is, the ink key settings are very close as the copy has not changed.
- the color change can be corrected by making the adjustment globally to all keys of each affected color. This color compensation will thus ensure that as subsequent newspaper editions are printed, the color complement for all advertisements will be comparable to that of the initial baseline newspaper edition.
- non-image areas are coated with water that prevents ink from adhering to the non-image areas and then to the printed material.
- the control of the ink water balance is extremely important as too much water or too little water greatly affects color.
- the initial adjustment of the ink keys is straight forward as the ink water balance is initially properly set up when the plates are new. This allows presetting of the ink keys based on densitometer readings (correlated RGB readings) as disclosed as the ink water balance will remain constant for some time.
- a more difficult problem that causes unacceptable color variations is changes in the ink water balance as the press warms up, as plates wear, and due to other factors which affect the ink water balance. These variations cannot be corrected by adjusting the ink keys but require an adjustment in the water metering system. There is at present no means for distinguishing the difference between density readings caused by ink film variations that can be corrected with adjustment of the ink key settings, and density readings caused by variations in ink water balance and can be corrected only with adjustment in the water metering system. For this reason automatic on-line systems that adjust ink keys have had only limited success, as they cannot be used for the more important on-line color variations that are caused by variations in the ink water balance .
- the following discussion discloses a method to determine the difference between an ink density difference reading that is due to a difference in ink film thickness that can be corrected with adjustment of ink keys and the same reading that is caused by a difference in ink water balance. This method may be implemented with advantageous results using the mark recognition system disclosed elsewhere in this disclosure.
- the disclosed method allows the system to be used for initial adjustment of the ink keys when new plates are installed, and for adjustment of either the ink keys or ink water balance as required to maintain consistent color during the run.
- the system employed to monitor or measure density of marks, areas, objects, etc. is illustrated in FIG. 77.
- RGB values can be read from anywhere on the entire repeat length or from calibration plates located on the side of the traverse.
- FIG. 82 illustrates an exemplary system process flowchart providing a procedure for automatic adjustment of ink keys based on the placement of individual marks within the range of a plethora of inking keys positioned across the plates that ink the web. This procedure (8200) will now be discussed in detail.
- the mark list is obtained from a register control process (8201) .
- the monitoring of marks for ink key adjustment consists of moving the camera to the calibration plate (8202) shown as (509, 510) of FIG. 5 on which is mounted chips of various colors including but not limited to yellow, magenta, cyan, black and various shades of white.
- chips are used for this process.
- These color chips (individually or collectively referred to as a calibration plate) have been read with a densitometer and the camera and both densitometer readings and their corresponding RGB values stored in a lookup table.
- Both densitometer reading and the corresponding RGB values of variations of the color of the calibration color chips are also stored in the look up table to relate the densitometer reading to and color variation that is obtained in RGB. Thus, for any RGB value the corresponding density reading can be determined from the look up table.
- the camera can be calibrated as often as necessary by moving the camera to the calibration plate and taking the RGB values of each color chip and comparing them with the RGB values stored in memory. Calibration of the camera is done either by mathematically correcting the values or by recalibration of the RGB values through the software provided for that purpose.
- the camera is then moved sequentially to each mark in the mark list of (8201) , starting with the first mark
- FIG. 83 illustrates an exemplary implementation of an ink/water balance control sequence (8300). This provides a means for measuring and correcting color variations due to changes in ink water balance. Since the applied water has an impact on the color of the web paper, the present invention makes use of the web corresponding to the plate gap to form a white reference that is used to then calibrate the water deposition rate in order to achieve desired color characteristics on the printed web.
- the camera used in this procedure may be calibrated using the procedure (8202) of FIG. 82.
- Marks are obtained from the mark list with their RGB values obtained and stored.
- the marks in this instance consists of the white areas around each mark set and the corresponding white areas of the plate/blanket gap directly opposite the mark set so as to have a comparison of the white area where there is printing and the white area where there is no printing.
- the typical ink/water balance adjustment process (8300) begins with obtaining a mark list (8301) and moving the camera/sensor to a plate/blanket gap on the web (8302).
- the web substrate density is obtained (8303) from the mark list corresponding to the ink keys.
- the substrate background density is then obtained from the plate/blanket gap (8304) to provide a reference for the values obtained from the mark list corresponding to the ink keys (8303) .
- the ink key values (8303) are compared to the plate/blanket gap values (8304) and if the difference is significant (8305) the ink/water balance is adjusted (8306) to improve the color ink deposition. Otherwise, the process is terminated (8307) . Note in a typical application, this process may be repeated as a background process to continuously monitor and correct the ink/water balance based on changes in press conditions.
- duplicate marks for every ink key and/or ink/water balance adjustment. These duplicate marks are be used in conjunction with corresponding color marks for each color ink key to inspect the color as applied to the web.
- a four-color press with 38 color adjustment keys would have 38 sets of duplicate marks and additionally 38 sets of color marks (yellow, cyan, magenta, for example) . If properly distributed throughout the web, these marks will permit the process of FIGs. 82 and 83 to completely control the color deposition and adjust for a wide variety of press variations and dynamic manufacturing conditions.
- FIG. 78 illustrates a flowchart representing how the present invention addresses the three major areas that define total quality in web offset printing in terms of a WEB OFFSET PRINTING Total Quality Control and Waste Reduction System (7801) .
- the Image Processor (7802) block may be viewed in some preferred embodiments as a representation of the hardware image processor depicted in FIG. 3 as computer (303) , touch screen control (302) , cameras (304, 308, 301) and in FIG. 1 as Computer (113), touch screen (111), color monitor (112) and cameras (107, 110) .
- the image processors obtain a number of different images (7803) as described within the teachings of the present invention from which information is extracted and utilized to improve quality and significantly reduce time and material waste in the three major areas of importance:
- the Preregistration (7804) sequence is broken into two areas :
- the Automatic Register Control (7805) sequence acts on normal register errors to automatically restore register to its previous position after tension and other disturbances during start-stops (7808) , splices (7811) , and any other condition that would produce a register error.
- the Color Monitoring and Control (7806) sequence provides the means as disclosed in FIG. 77 to measure color variations. With two different variations of this concept one with application in the initial adjustment of ink keys when setting up a new job, and the other for monitoring color through the run to maintain consistent quality. These two methods use the an exemplary process flow as illustrated in FIG. 77 and are fully disclosed previously in this disclosure.
- each of the error correction paths (7804, 7805, 7806) may be activated asynchronously in many implementations to dynamically adjust printing operation prior to and during normal running operation (7813) .
- this functionality may be easily implemented in many multi-tasking or threaded task processor environments.
- FIG. 78 The system/process flow illustrated in FIG. 78 is unique in the industry for several reasons. Generally, most of the features present in this automated total quality control system are currently performed manually in the prior art, and in situations where they are not, the systems tend to be single-purpose in their functionality. The present invention breaks with tradition by permitting ALL the errors associated with poor quality and printing waste to be attacked simultaneously, thus providing for a superior web product as well as drastically reduced waste from the overall printing process.
- the present invention should be viewed in terms of the disclosed system, method, process, and additionally in terms of the product created by the process disclosed herein.
- the rationale for the claim of the printed web itself as being novel lies in the fact that the present invention web printing systems and associated method/processes have a product distribution that is of a different kind and substantially higher quality than that possible with conventional web printing systems.
- an analysis of the comparative quality distributions of the present art (8010) versus the present invention (8020) based on the exemplary economic analysis presented in FIG. 19 and other data reveals that the distribution curve for the present invention is markedly different than that possible with the prior art.
- the distribution indicates that while the majority (50%) of the product produced is of a poor register quality (8011) that is nonetheless marginally acceptable for the customer, there typically exists approximately 10% of the printed product that is unacceptable (8012), yet still shipped to the customer for economic reasons.
- up to 10% of the printed material is considered waste/scrap (8013) due to setup registration errors. Note that this percentage will be higher for short runs, but even for the example illustrated in FIG. 19 is approximately 2% for a common newspaper with a moderate production run.
- the present invention product distribution (8020) is of a completely different character.
- the majority (90%) of the product produced is of the high-quality (high registration compliance) variety (8021)
- there is negligible waste (8022) and only a small quantity of product (8%) that is of the acceptable/marginal variety (8023) .
- the acceptable/marginal (8023) component of the product output represents the sum of uncorrectable errors in the printing process plus the feedback control loop delay associated with the present invention as implemented on the printing press. Since the printing control system and method described herein permits the observation camera or other sensor to be placed directly after application of the ink to the web material , this feedback control loop may be much shorter than possible with the prior art.
- FIG. 19 While the example provided in FIG. 19, and the distribution charts illustrated in FIG. 80 are only exemplary, they indicate that the present invention when applied to web printing permits a given lot of web material to be qualitatively different than that produced by the prior art. Whereas the prior art printing methods were forced from an economic perspective to accept marginal color registration in many circumstances (because to do otherwise would increase waste/scrap quantities that would drive up the cost of the print job), the present invention changes the printing dynamic by essentially eliminating waste completely from the printing process , and converting what was waste into acceptable/marginal output product.
- the vast majority of the product produced by the present system is of the high-quality superior registration quality, and as such the disclosed invention produces a product that is consistently of a different kind than that possible with the prior art. As such, it should be considered a ⁇ super-registered' web product, consistently of better quality than that possible with the prior art. While this super-registered web product has a direct economic impact on reducing waste and increasing profits for a given printer, it should be noted that the present invention also has the capability of permitting older printing presses that have a plethora of alignment and aging problems to produce a super-registered web product in situations where the printing press would not have been able to produce even acceptable/marginal product in the past.
- the dynamic feedback provided by the present invention in conjunction with a short feedback loop and three-dimensional compensation for web and press variations permits even an aging press in many circumstances to produce an overall product distribution that surpasses the capabilities of even newer modern day printing presses.
- FIGs. 1-80 may be reduced to computer instruction codes and embodied on a computer readable storage means. This may take the form of a wide variety of storage media well known in the art and/or contemplated for future use. Thus, the present invention specifically anticipates the incorporation of the system and methods discussed herein in the form of tangible computer software products.
- the present invention specifically anticipates that one or more components of the present invention may be implemented using the MicrosoftTM WindowsTM operating environment in all its variations or its equivalent commercial embodiments, including but not limited to any system incorporating a graphical user interface.
- the present invention has wide application to all forms of the printing industry, as well as other industries that operate on webbed products using rollers, printing, punching, and the like. Since a primary advantage of the present invention over the prior art is the simultaneous minimization of setup time and resulting waste products, it can easily be understood by one skilled in the art that the present invention has the potential to substantially improve the environment by reducing landfill waste and the like.
- FIG. 19 An exemplary waste savings analysis is illustrated in FIG. 19, wherein an exemplary newspaper application (1900) consisting of a circulation of 50000 newspapers (1901) using standard printing press configurations (1902, 1903, 1904, 1905, 1906) and conventional ink and paper costs (1907, 1908, 1909) . Standard production run information (1910, 1911, 1912, 1913) results in a savings of approximately 1000 papers per make ready count (1914) .
- Equating this lost newsprint to manufacturing cost in terms of ink and paper cost yields a savings of US$2,487 per week or US$129,336 per year. What is significant to note in this analysis is that the present invention installed cost may in fact be less than this yearly material savings.
- the printer has an economic incentive to reduce waste by making use of the present invention, as the equipment costs can be recovered within one year, with subsequent years having increased profit margins associated with annual savings of US$129,336 throughout the useful life of the present invention. Additionally, a total of 235 tons/year of unnecessary landfill waste is eliminated via utilization of the present invention on top of the yearly cost savings.
- the operation of the present invention may be affected both locally and remotely. While one skilled in the art will clearly recognize that the present invention may be integrated into a printing press environment as illustrated by the examples in FIGs. 1-24, there also exists a possibility of using the system as described in a remote capacity.
- FIG. 81 the general system configuration of FIG. 57 has been augmented by identification of the local mark recognition system (8100) comprising an image acquisition means (8103) , image processor means (8104) , operator interface display means (8105) , and web press motor control means (8106) .
- This mark recognition system (8100) operates as described previously in regards to inspecting a web material (8101) on which register marks (8102) have been applied.
- This local system (8100) is augmented for remote applications with a communication network means (8106) connected to a remote processing system means (8107) , which is typically a personal computer running a graphical operating system such as a variant of the Microsoft Windows operating environment but is not limited to this configuration.
- This remote processing system means (8107) has associated with it a remote storage device database means (8108) containing register mark software and/or other maintenance and inspection software.
- the mark recognition system (8100) may be remotely updated and/or inspected from a location not local to the actual printing press. This capability is extremely useful in both troubleshooting printing press problems but also provides a means for providing custom software updates to a printing press system based on demands of various new printing requirements.
- FIG. 81 may conceivably be any form of digital and/or analog communications medium, in many instances the preferred method of communication will make use of the Internet. This technique in some circumstances will also permit the mark recognition system (8100) to download updated register mark software from a remote database (8108) via use of the remote processing system (8107) .
- the present invention may be incorporated into a wide variety of system contexts, although one preferred system context in regards to web printing is illustrated by the configurations in FIGs. 1-29. Note, however, that these are only exemplary system applications. Given the relative low cost of this mark recognition system, it is amenable to a wide variety of other applications where a low cost alternative to Zoom Lens Calibration is desirable.
- the presently disclosed system may in some circumstances be augmented by a Zoom Lens Calibration system to affect a system with hybrid capabilities and costs of each system.
- the present system teaches a multifunctional use of a camera system, including inspection, registration (both rough initial and fine) , continuous press adjustment, color monitoring, color adjustment, ink key modulation, ink/water balancing, and/or remote press diagnosis and control. Individually most of these features are lacking in the prior art, and when integrated into a single system they represent a significant cost reduction over other single-function systems.
- the present invention actually SAVES the printer money by significantly reducing waste during preregistration and during normal press operation. This waste savings can in some cases actually pay for the added equipment costs associated with the present invention and thus provide an economic incentive for the printer to conserve environmental resources by not wasting paper and other consumables.
- the present invention for the first time permits in situ calibration of color ink deposition using the web itself and/or external calibration plates to automatically adjust registration, color-to-color register, color quality, ink key adjustments, as well as ink/water balancing. All of these adjustments traditionally were manually controlled, and the use of the present invention teachings permits a significant reduction in manual labor and maintenance associated with traditional printing processes.
- the ability to remotely monitor and diagnose a printing press operation using the present invention is yet another feature that promotes the economics of this new paradigm, as the prior art does not teach this functionality in the context of register mark or printing press processing.
- the present invention permits a product mix of a different and higher quality kind to be produced by the average printing press, thus both reducing waste and allowing older presses to actually produce product that is not generally possible using manual control techniques.
- retrofits of existing presses can fit them to be economically viable in short run situations where previously the waste created was a significant barrier to their economic viability.
- the present invention permits total quality management of the printing process, and while the general commercial printing industry is only one area in which the teachings of the present invention may be applied, it is one significant area in which the present invention can significantly change both the type of product produced as well as the way in which it is manufactured.
Landscapes
- Inking, Control Or Cleaning Of Printing Machines (AREA)
Applications Claiming Priority (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US10545698P | 1998-10-23 | 1998-10-23 | |
US105456P | 1998-10-23 | ||
PCT/US1999/024045 WO2000027638A1 (en) | 1998-10-23 | 1999-10-22 | System and method for register mark recognition |
Publications (1)
Publication Number | Publication Date |
---|---|
EP1123209A1 true EP1123209A1 (de) | 2001-08-16 |
Family
ID=22305955
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
EP99971761A Withdrawn EP1123209A1 (de) | 1998-10-23 | 1999-10-22 | Vorrichtung und verfahren zum registrieren von registermarken |
Country Status (3)
Country | Link |
---|---|
EP (1) | EP1123209A1 (de) |
AU (1) | AU1205400A (de) |
WO (1) | WO2000027638A1 (de) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2011110885A (ja) * | 2009-11-30 | 2011-06-09 | Komori Corp | 印刷機の見当合わせ装置及び印刷機の見当合わせ方法 |
CN110893725B (zh) | 2018-09-12 | 2021-08-17 | 海德堡印刷机械股份公司 | 具有圆形测量标记的对版-套准测量 |
CN111823699B (zh) * | 2020-06-09 | 2021-11-30 | 宁波欣达印刷机器有限公司 | 一种规避凹版印刷机因换料导致套色误调的方法及*** |
IT202000007145U1 (it) * | 2020-12-29 | 2022-06-29 | Fedrigoni Spa | Sistema di monitoraggio per impianti per la produzione di carta. |
Family Cites Families (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4546700A (en) * | 1981-12-30 | 1985-10-15 | Kollmorgen Technologies Corporation | Method and apparatus for sensing and maintaining color registration |
DE3809941A1 (de) * | 1987-03-26 | 1988-10-06 | Koenig & Bauer Ag | Verfahren zum positionieren von plattenzylindern in einer mehrfarben-rotationsdruckmaschine |
DK164972C (da) * | 1989-06-08 | 1993-02-15 | Bjarne Chr Nielsen | Fremgangsmaade og anlaeg til kontrol af trykkeprocesser |
US5828075A (en) * | 1996-10-11 | 1998-10-27 | Hurletron, Incorporated | Apparatus for scanning colored registration marks |
NL1004663C2 (nl) * | 1996-12-02 | 1998-06-03 | Q I Press Controls V O F | Werkwijze en inrichting voor het controleren van drukwerk. |
US5809894A (en) * | 1997-02-20 | 1998-09-22 | Advanced Vision Technology, Ltd. | System and method for registration control on-press during press set-up and printing |
CA2302197C (en) * | 1997-09-02 | 2006-05-16 | Innolutions, Inc. | Method and apparatus for register mark identification |
-
1999
- 1999-10-22 EP EP99971761A patent/EP1123209A1/de not_active Withdrawn
- 1999-10-22 AU AU12054/00A patent/AU1205400A/en not_active Abandoned
- 1999-10-22 WO PCT/US1999/024045 patent/WO2000027638A1/en not_active Application Discontinuation
Non-Patent Citations (1)
Title |
---|
See references of WO0027638A1 * |
Also Published As
Publication number | Publication date |
---|---|
WO2000027638A1 (en) | 2000-05-18 |
AU1205400A (en) | 2000-05-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US6874420B2 (en) | System and method for register mark recognition | |
EP0884181B1 (de) | Verfahren zur Regelung von Betriebsvorgängen einer drucktechnischen Maschine | |
CA2078930C (en) | Register mark | |
US6031932A (en) | Automatic inspection of printing plates or cylinders | |
US8237828B2 (en) | Device for inspection of print products | |
JP7150530B2 (ja) | 画像センサの自動的な較正 | |
FI95888B (fi) | Menetelmä painatuksen laadun valvomiseksi | |
AU2017222088B2 (en) | CNC flatbed cutting machine, its method of operation, and a graphics sheet with a fiducial that indicates the orientation of the graphics sheet | |
DE10215511A1 (de) | Verfahren und Einrichtung zur Messung von Positionen von durchlaufenden Bogen | |
US20100039510A1 (en) | Method and DEVICE for PRINT INSPECTION | |
CN113195235A (zh) | 用于定位玻璃支撑件的方法/设备和用于在所述玻璃支撑件上打印的包括用于定位的所述方法/设备的方法/*** | |
JP7350637B2 (ja) | 画像検査のための高速画像歪み補正 | |
JP2011076204A (ja) | 印刷物検査方法及び印刷物検査装置 | |
WO2000027638A1 (en) | System and method for register mark recognition | |
CN108701358B (zh) | 用于定位印刷基底的方法/装置以及包括用于定位的所述方法/装置的印刷方法/*** | |
US20200079119A1 (en) | Method for automated alignment and register measurement using circular measuring marks | |
EP1531611B1 (de) | Vorrichtung und Verfahren zur Messung der Längenänderung der Vorschubspindel in einem Belichter für Druckvorlagen | |
KR100421266B1 (ko) | 광전자장치에의한자동평가방법 | |
JPH08276569A (ja) | 紙面上の画像の印刷品質を自動検査するために使用することが意図された基準モデルを作成するための方法 | |
JP3207931B2 (ja) | 画像検査装置 | |
JP6665903B2 (ja) | 特徴画像生成装置、検査装置、特徴画像生成方法及び特徴画像生成プログラム | |
DE102007041673A1 (de) | Verfahren und Messeinrichtung zur Gewinnung von Druckbild- oder Druckplatteninformationen | |
US9531936B2 (en) | Camera system, colour measuring system, and offset printing press | |
EP0437648A1 (de) | Verfahren zur Qualitätsbeurteilung von Druckerzeugnissen | |
JP2007327953A (ja) | センサ装置 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PUAI | Public reference made under article 153(3) epc to a published international application that has entered the european phase |
Free format text: ORIGINAL CODE: 0009012 |
|
17P | Request for examination filed |
Effective date: 20001222 |
|
AK | Designated contracting states |
Kind code of ref document: A1 Designated state(s): AT BE CH CY DE DK ES FI FR GB GR IE IT LI LU MC NL PT SE |
|
STAA | Information on the status of an ep patent application or granted ep patent |
Free format text: STATUS: THE APPLICATION IS DEEMED TO BE WITHDRAWN |
|
18D | Application deemed to be withdrawn |
Effective date: 20040504 |