GB2224831A

GB2224831A - System for processing line scan video image signal information

Info

Publication number: GB2224831A
Application number: GB8921366A
Authority: GB
Inventors: Milan Fuchs
Original assignee: Radix Systems Ltd
Current assignee: Radix Systems Ltd
Priority date: 1988-09-21
Filing date: 1989-09-21
Publication date: 1990-05-16
Anticipated expiration: 2009-09-21
Also published as: GB8822183D0; GB8921366D0; GB2224831B

Abstract

The speed of processing is increased by providing a plurality of image processors 16 each of which processes concurrently a different package of the image signal information derived from the video signal. Each image processor has an input buffer set up to recognise and store only information derived from pixels in a group the numbers of which it has in memory. The signal information supplied to the image processors is simplified by a video signal processor 11 which compares the individual pixel signals in the video signal input with computer-controlled threshold values and outputs to an image information bus only signals representing the occurrences and directions of threshold crossings and the numbers of the pixels concerned. The system is especially for the detection of faults in travelling materials or objects. <IMAGE>

Description

METHOD AND APPARATUS FOR PROCESSING LINE SCAN OPTICAL IMAGES This invention relates to a method and apparatus for the processing of a line scan optical image to detect imperfections or blemishes in a travelling sheet or to determine undesirable objects within a stream of objects or particulate material, generally for the purposes of rejecting such objects from the stream of acceptable objects or material. The invention is especially suitable for processing optical image information from photo detector arrays arranged in a single line (linescan CCD arrays), where high speeds and correspondingly high material handling capacities are required.

The method and apparatus of the present invention is typically applicable to the sorting and/or rejection of incorrectly sized or shaped or coloured pharmaceutical capsules or tablets; of confectionery such as sweets; or particulate plastics such as glass fibre reinforced pellets or pellets of different colours; of foodstuffs such as breakfast cereals, snack foods, rice and other granular or particulate materials; or to the detection of blemishes in continuous sheet such as plastics sheet.

It is generally known [ e.g. Automated Visual Inspection, D.G. Batchelor et al, IFS (Publications) Limited 1985 ] that line scan cameras can be used for size or defect inspections, especially of moving objects. It is also known that the capacity of inspection machinery in some applications is limited by the processing capability of current technology electronic processing elements such as microprocessors and transputers. It is therefore an object of the present invention to improve that processing capability.

According to the invention, there is provided a method of processing a train of video image signal information derived from a video signal obtained by at least one linescan operation, especially for the detection of faults or inconsistencies in a travelling sheet of material or in objects or particulate material travelling in a stream, wherein the image signal information is allocated into packets (which may or may not overlap) and the individual packets of image information are processed concurrently in different image processors. In, for instance, an object or particle rejection system, in which a linear array of detectors scans a travelling stream of objects or particulate material, higher speeds are possible using the invention than are achievable with the existing systems.

The invention further provides apparatus for processing video signal information obtained from at least one line scan camera, especially for the detection of faults or inconsistencies in a travelling sheet of material or in objects or particulate material travelling in a stream, comprising a video signal processor deriving a train of video image signal information from the camera output, and a plurality of image processors all receiving said image signal information but each arranged to recognise and process only a respective packet of that information, whereby different packets of the image signal information (which may or may not overlap) are processed concurrently.

Arrangements according to the invention will now be described by way of example and with reference to the accompanying drawings, in which: Figure 1 is a schematic block diagram of a system operating according to the invention, Figure 2 shows in more detail a video processor of Figure 1, and Figure 3 shows in more detail an image processor of Figure 1.

Referring firstly to Figure 1, the video signal information is obtained from one or more line arrays of photo detectors. Each detector array is built into a line scan camera 10 which can have a resolution of typically 2048 elements, but higher or lower resolution arrays may be employed. The video signal output from each camera is delivered to a video processing unit 11.

The output of each camera 10 during an integration or exposure cycle time, i.e. a single scan, comprises a respective signal (pixel signal) from every detector in the array having a magnitude proportional to the light received by that detector. These pixel signals are available serially, the magnitude of the signal of one pixel being followed by the magnitude of the signal of the next succeeding pixel in the line scanned by the detector array, each lasting a predetermined period of time.

The video processing unit 11 includes a clock signal generator 27, random access memories (RAMs) 28,29, an analogue-to-digital converter 30, a video signal comparator 31 and a digital-to-analogue converter 32.

The serial signal received under clock control by the video processing unit 11 from any one camera is compared against one or more thresholds provided by reference voltages which are preset depending on the material being scanned, the lighting, the type of feature to be detected and so on. Whenever a signal crosses the applicable threshold in either direction an "edge" signal is generated by an edge detector. The threshold level can be constant, i.e. identical for all pixel signals, or variable as required. A variable threshold will allow for presetting of a different threshold level for each pixel, or for different groups of pixels, typically to compensate for non-linear lighting or gradual variations in ambient conditions.

The camera video signal is in analogue form, normally a voltage produced for each pixel corresponding in magnitude to the amount of light received by the detector element. All line scan cameras connected to the system will typically be synchronised to the same clock information and therefore the same number pixel signal will be available at the same time for corresponding pixels scanned by all detector arrays.

Figure 2 shows the video processor 11 receiving video or pixel signals from cameras A and B simultaneously. In normal operation, the analogue video signal inputs from cameras A and B are compared with the appropriate threshold or reference voltage in the video comparator 31 which includes four edge detector circuits, that is to say two for camera A to detect threshold crossings by each pixel signal and likewise two for camera B. The threshold crossing or edge signal outputs for cameras A and B are denoted TAl, TA2 and TB1, TB2, respectively. A fifth output PSTB provides a strobe pulse whenever there is an edge signal output from any of the four edge detectors; and all five signals are fed to a camera bus 15.

The reference voltages setting the thresholds are input to the video processor in digital form from the central computer bus 22 via an interface 12 to the RAM 28, whence they are supplied by the digital-to-analogue converter 32 to the edge detector circuits.

The analogue-to-digital converter 30 supplies digital conversions of the camera video signals to the single scan RAM 29 in an off-line auto-calibration operation for obtaining information on, for example, the background lighting profile so as to enable a central processor unit (CPU) 20 to introduce appropriate compensation when defining the reference voltages to establish the thresholds for the edge detectors in the video comparator 31.

The clock generator circuit 27 not only generates appropriate clock pulses for the two cameras but also counts the pixel numbers, the count being used for the following functions: - to identify the appropriate address in the single scan RAM 29 for each analogue-to-digital converted pixel signal strength, - to identify the appropriate address in the threshold RAM 28 so that the correct (current) reference voltage can be given to the edge detectors, - to supply the current pixel number to the camera bus 15 for combination with the threshold states and strobe signals.

Depending on the objects being viewed and the processing speed requirements, the image signals on the camera bus 15 are supplied to one or more of the image processors 16 (Fig. 1). Typically, but not necessarily, several image processors 16 will be employed for any stream of objects being presented within the field of view. Each image processor is preset to select only a sector of information which is of interest to that particular image processor. A sector of information may comprise a predetermined number of pixel signals from the total number available and may or may not overlap with adjacent sectors being processed by other processors.

It is therefore possible for each image processor to be simultaneously processing a part only of the total information. That part may represent a small portion within the total field of view, or the ability is provided to process different features within the same field of view. The image signals on the camera bus 15 only provide information on the current pixel number, whether a threshold-crossing transition has occurred (the strobe signal) and the state, i.e. above threshold level or below threshold level, of the video signal from each of the cameras A and B. Since this information is of a simple character it can be processed very fast within an image processor; it is, nevertheless, quite adequate for the kind of defect detection for which the system is intended.

Each image processor 16 includes an asynchronous buffer 17 capable of holding multiple pixel state information and the specific information which the particular processor is preset to handle is stored within this buffer. Only the states of the pixels appertaining to the sector of interest to the image processor and whereof a transition (edge) has occurred are stored in this buffer.

Referring to Figure 3, the pixel range appropriate to the sector of interest of the image processor is input from the CPU 20 via the bus 22 and a passive interface 33 to store in a sector select RAM 34. Associated sector select logic 35 reads the pixel number for each edge event of the incoming image signal information from the camera bus 15 and, if and only if that pixel number is in the range stored in its RAM 34, the select logic passes the respective strobe signal PSTB for application to a storage register (FIFO) 36. Simultaneously, the respective pixel number and state are fed to the FIFO 36 and are consequently strobed into temporary storage there. Thus, the only information records stored in the FIFO 36 are the pixel numbers and states TA1, TA2, TB1, TB2 of pixels within the sector of interest that show edge events.

An image processor CPU 18, e.g. a microprocessor, can read the information from the FIFO 36, via an internal image processor bus 37, and process the desired feature or features. Such features can be the width of an object scanned, its length, area, volume, shape, colour, transmittance, and so on. For the processing of two or more dimensions or parameters (area, shape, colour), additional information may be required by the processing element. This may come in the form of a second or subsequent camera image signal or from external inputs directly available to the CPU 18, as at input port 19, via the bus 37. The use of multiple line scan cameras, e.g.

differently orientated, can derive two or three dimensional information for processing. Colour filters can be employed within the optical systems of the line scan cameras to obtain information on colour or colour defects regarding the objects scanned. The direct external inputs typically provide timing references giving information as to, for instance, the speed of travel of objects being viewed or the current location of an object to be rejected. Each image processor 16 also has, at 38, a direct output drive capability so that single or multi-channel reject mechanisms can be operated, alarms raised, or other appropriate action taken.

The image processor CPU 18 constantly examines the condition of the FIFO 36 and, if it is not empty, will process each pixel number/edge detector state record in turn in order to determine the parameters required by the program resident in an EPROM 39 whilst taking into account parameter values which have been down-loaded from the main CPU 20 via the bus 22, passive interface 33 and a dual port RAM 40. From time to time the image processor CPU 18 will deposit information on its own performance and the characteristics of the objects being viewed in the dual port RAM 40 for subsequent collection by the main CPU 20.

Provision is also made for interfacing by a human operator or with a host computer. This can take place through the main CPU 20 which is capable of accepting information from an operator peripheral such as a keyboard 26 via a peripheral interface 41 and passing selected information on via the computer bus 22 to random access memories 23 of the individual image processors 16 and/or the video processing unit 11. Typically the information passed to the video processor 11 will be the threshold level(s) or lighting profile. The information passed to the image processors 16 may be the definition of sectors (areas of interest), type of processing, e.g. calculate area of the object scanned, acceptability levels and engineering parameters. The main CPU 20 may also be interfaced to other peripheral devices such as a display monitor 24, a hard copy printer 21, and so on, and have connection capability to other systems, such as the host computer already mentioned for overall information management.

Claims

CLAIMS:

1. A method of processing a train of video image signal information derived from a video signal obtained by at least one line-scan operation, especially for the detection of faults or inconsistencies in a travelling sheet of material or in objects or particulate material travelling in a stream, wherein the image signal information is allocated into packets (which may or may not overlap) and the individual packets of image information are processed concurrently in different image processors.

2. A method according to Claim 1, wherein the video signal comprises at least one train of individual pixel signals from respective detectors in a line scan array, and each said packet of image information contains the image signal information derived from a selected group of pixel signals.

3. A method according to Claim 1 or Claim 2, wherein the video signal is obtained from a plurality of line scan operations providing a plurality of trains of pixel signals which are processed substantially concurrently to derive the image signal information that is then allocated into packets.

4. A method according to any preceding claim, wherein the processing of the video signal to derive the image signal information for allocation into packets includes comparison of individual pixel signals with respective threshold levels, the image signal information including the pixel number of any pixel signal in respect of which an edge event has occurred by crossing of the respective threshold level and an indication of the state achieved consequent on the direction of crossing.

5. A method according to any preceding claim, wherein, aside from the derivation of the image signal information, the video signal is separately monitored off-line, to achieve compensation for background or environmental changes.

6 Apparatus for processing video signal information obtained from at least one line scan camera, especially for the detection of faults or inconsistencies in a travelling sheet of material or in objects or particulate material travelling in a stream, comprising a video signal processor deriving a train of video image signal information from the camera output, and a plurality of image processors all receiving said image signal information but each arranged to recognise and process only a respective packet of that information, whereby different packets of the image signal information (which may or may not overlap) are processed concurrently.

7. Apparatus according to Claim 6, wherein the video signal comprises at least one train of individual pixel signals from respective detectors in a line scan array, and each image processor has an input buffer that is programmed to recognise and store for processing only the image signal information derived from a selected group of said pixel signals.

8. Apparatus according to Claim 7, wherein the video signal processor includes comparator means for comparing individual pixel signals with respective threshold levels provided by reference signals and generating image information signals indicating when a pixel signal has crossed a respective threshold level and in which direction, and a clock generator establishing a count whereby each pixel signal is identified by a pixel number, the pixel number of any pixel signal that has crossed a respective threshold also appearing in the image signal information.

9. Apparatus according to Claim 8, wherein the video signal information to the video signal processor comprises a plurality of trains of pixel signals from a plurality of line scan detector arrays, and the video signal processor is arranged to process the incoming pixel signals in pairs or groups and to generate a strobe signal that appears in the image signal information if any pixel signal in a pair or group has crossed a respective threshold level.

10. Apparatus according to Claim 9, wherein the input buffer of each image processor is supplied with the numbers of the pixel signals in its selected group by a main central processing unit and is arranged to store image signal information appertaining only to pixel signals that both have numbers appearing in that selected group and are accompanied by strobe signals.

11. Apparatus according to any of Claims 7 to 10, wherein each image processor includes a respective processing unit that operates on the image signal information obtained from storage in the input buffer in accordance with a programme also stored in the image processor and/or external input signals applied directly to that image processor.

12. Apparatus according to Claim 11, wherein each image processor is also arranged to directly output drive signals under the control of its processing unit.

13. Apparatus according to Claim 10, or Claims 10 and 11, or Claims 10 and 12, wherein the video signal processor further comprises monitoring means for monitoring the incoming video signal information to provide the main central processing unit with data enabling it to compensate for background or environmental changes.

14. Apparatus according to Claim 10, or Claims 10 and 11, or Claims 10 and 12, or Claim 13, wherein the threshold signals in the video signal processor, the pixel group selection by the input buffer of each image processor, and the stored programme in each image processor, can be modified or changed by the main central processing unit automatically and/or in response to human operator input.

15. Apparatus for processing line scan video signal information, substantially as described with reference to the accompanying drawings.