WO2019007405A1 - 一种斜射式锯材缺陷检测装置及检测方法 - Google Patents

一种斜射式锯材缺陷检测装置及检测方法 Download PDF

Info

Publication number
WO2019007405A1
WO2019007405A1 PCT/CN2018/094708 CN2018094708W WO2019007405A1 WO 2019007405 A1 WO2019007405 A1 WO 2019007405A1 CN 2018094708 W CN2018094708 W CN 2018094708W WO 2019007405 A1 WO2019007405 A1 WO 2019007405A1
Authority
WO
WIPO (PCT)
Prior art keywords
thickness
sawn
sawn timber
industrial camera
tested
Prior art date
Application number
PCT/CN2018/094708
Other languages
English (en)
French (fr)
Inventor
徐兆军
梅长彤
倪申建
朱南峰
周海燕
张攀
刘斌
王继刚
郝淼
万智龙
马辉贤
余晓
Original Assignee
南京林业大学
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Priority claimed from CN201710545226.3A external-priority patent/CN107388973A/zh
Priority claimed from CN201710545221.0A external-priority patent/CN107388962A/zh
Priority claimed from CN201710545229.7A external-priority patent/CN107561091A/zh
Application filed by 南京林业大学 filed Critical 南京林业大学
Publication of WO2019007405A1 publication Critical patent/WO2019007405A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
    • G01B11/00—Measuring arrangements characterised by the use of optical techniques
    • G01B11/02—Measuring arrangements characterised by the use of optical techniques for measuring length, width or thickness
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
    • G01N21/84—Systems specially adapted for particular applications
    • G01N21/88—Investigating the presence of flaws or contamination
    • G01N21/89—Investigating the presence of flaws or contamination in moving material, e.g. running paper or textiles

Definitions

  • the invention relates to a detecting device and a detecting method, in particular to a detecting device for an oblique sawing material defect and a detecting method thereof, and belongs to the technical field of measurement.
  • the detector uses visual methods to determine the type of defect.
  • a caliper or other gauge is used to determine the size of the crack defect.
  • the degree of cracking of the sheet is determined according to the width and size of the crack. Not only the workload is large, it is easy to cause visual fatigue, and the measurement process is affected by subjective factors of the tester.
  • the efficiency and quality of the test are difficult to guarantee, and it is difficult to adapt to large-scale industrial automation production, especially as consumers demand product quality. Improvements, relying on manual inspection and measurement, are increasingly unable to meet the requirements of today's industrial fields.
  • the blunt edge phenomenon in processing defects when the wood products are processed and produced, the logs are sawn into sheets of a certain thickness.
  • the cross section of the logs can be approximated as a circle, and the logs can be approximated into a cylindrical shape.
  • the general sawing process It is parallel to the cylinder axis.
  • the sawn long section of the board is not a regular rectangle, but the two sides have irregular curved trapezoids.
  • Figure 1 is a schematic view of the structure of a blunt plate, as shown in Figure 1: This plate is called a blunt edge plate, and the outer surface of the blunt edge sawn timber is divided into two upper and lower wide faces and two left and right narrow faces, since the logs themselves are not A regular cylinder with defects such as branches and bends makes the narrow surface of the blunt-edge sawn timber have an irregular shape, and the irregular narrow-faced surface needs to be removed during processing.
  • 2 is a schematic structural view of a blunt edge plate in a sawing state, as shown in FIG. 2: generally, in the subsequent sections, the blunt edge of the blunt edge plate is often removed, and if the narrow surface is removed, the blunt edge defect is formed, affecting The grade and use of sawn timber will be wasted if removed too much.
  • the distance from the wide face to the nucleus of the blunt-edge sawn timber it is divided into the inner material surface and the outer material surface, the inner material surface is close to the inner core, and the outer surface is far.
  • experienced workers are required to visualize the outer surface of each sawn timber. With their own experience, the manual operation of the workers seriously affects the processing precision and efficiency, and causes the resources of the blunt edge sawn timber. waste.
  • wood panels or wood products will shrink and swell as the relative humidity of the surrounding environment changes. Since wood is a kind of heterogeneous anisotropic material, dry shrinkage and swelling are also anisotropic. Wood panels placed in natural environment will warp, and woodboards in different temperature and humidity environments will have different warpage. .
  • Figures 3 to 4 are graphs of warpage measurement.
  • the floor surface is placed up on the horizontal test bench surface, and the knife ruler or steel ruler is placed vertically against the floor.
  • the knife ruler or steel ruler is placed vertically against the floor.
  • the ratio of the maximum chord height h max to the measured width ⁇ of the floor is the warp in the width direction.
  • f ⁇ is expressed as a percentage, accurate to 0.01%.
  • the laser scanning method uses the feature of the sawn material to be associated with its surface contour, and the thickness of the portion where the defect is located is thinner than the thickness of the normal sawing material.
  • the specific step is that the laser displacement sensor emits a light source to the surface of the test piece fixed on the workbench, and then the laser displacement sensor receives the reflected light on the surface of the test piece, and the thickness profile information of the test point can be obtained according to the length of the projected and reflected light path.
  • the laser displacement sensor is movable in the horizontal direction so that the thickness profile information of the entire test piece surface can be scanned.
  • the thickness profile is converted into a thickness profile image by a dedicated processing software, and the defect is located and identified through a series of image processing and feature recognition.
  • This method can accurately measure the defect information at the measurement point, but it also has shortcomings, such as slow detection speed. In the actual detection, a large number of mapping points are needed to obtain the wood defect contour information, so this method cannot be adapted.
  • the laser scanning method is a laser time difference type, which emits a laser pulse, measures the time difference from the emission to the return of the pulse, and converts the distance from the probe to the measured object. Because the light travels quickly, accurate measurement When the small displacement is small, the accuracy of the time measurement is very high, otherwise the accuracy of the measurement is difficult to guarantee, and the price of the sensor is also high.
  • an object of the present invention is to provide an oblique sawing material defect detecting device which is safe and reliable, has high measurement accuracy, and is convenient to use.
  • An oblique sawing material defect detecting device comprises a conveying table, a laser emitter, a bracket and an array industrial camera, and the bracket bridge is mounted on the conveying table, and the laser emitter is arranged on the bracket in turn along the conveying direction of the conveying table.
  • An array of industrial cameras, and the light source emitted from the laser emitter to the surface of the sawn timber on the conveyor table is reflected by the sawn timber and then received by the array industrial camera, and the array industrial camera is simultaneously connected with a computer.
  • the incident laser line emitted by the laser emitter is at an angle of 90° to the optical axis of the area array industrial camera, and the incident light is oblique, and 0 ⁇ incident angle ⁇ 90°.
  • the saw material defect described above is any one of a sawing material growth defect, a sawing material processing defect, and a sawn material drying defect.
  • the light source emitted from the laser emitter to the surface of the sawn material on the conveyor table is a fan-shaped laser light source, so that the surface of the sawn material is formed into a light strip, and the light strip is a strip-shaped light strip.
  • the present invention also provides a method of detecting a sawing material defect based on the above-described oblique sawing material defect detecting device.
  • An oblique sawing material defect detecting method comprises the following steps:
  • the laser emitter emits a laser light source to the sawn material passing through the conveyor table, and is reflected by the surface of the sawn timber after being reflected by the surface array industrial camera, and the incident laser line and the area array industrial camera emitted by the laser emitter are The optical axis is at an angle of 90°, and the incident light is oblique, and 0 ⁇ incident angle ⁇ 90°;
  • the area array industrial camera converts the received photoelectric signal into a digital signal and transmits it to the computer.
  • the data acquisition card in the computer receives the digital signal and is processed by the computer processing software to obtain the measured cross section of the sawn timber in the thickness and The contour image information in the width direction, and then obtaining the thickness value pixel and the width value pixel of the sawn material according to the obtained contour image information, and converting the pixel unit into a length unit;
  • step S3 according to the contour information of the measured cross section of the sawn timber obtained in step S2 and the conversion relationship between the pixel unit and the length unit, further obtain the position information of the sawn material defect, and perform corresponding removal according to the position information of the sawn timber defect. Defect operation.
  • x is the detected thickness value of the sawn timber to be tested
  • y is the offset of the surface of the sawn timber to be tested in the area array industrial camera
  • y is expressed as the thickness of the sawn timber in the contour image
  • k is the spatial resolution in the thickness direction
  • k is a linear relationship coefficient between the offset y of the surface of the sawn material to be measured and the reference bottom surface of the sawn timber in the area array industrial camera.
  • ⁇ is the incident angle of the laser emitter
  • ⁇ is the angle between the optical axis of the area array industrial camera and the normal to the surface of the sawn material
  • f is the camera focal length
  • L is the measured thickness when the thickness is 0.
  • the height of the table is determined, and L, f, and ⁇ are all fixed values, and k is also a fixed value.
  • ⁇ is the actual width value of the sawn timber to be tested
  • k2' is the spatial resolution of the width of the sawn timber to be tested
  • W' is the width value of the sawn timber to be tested in the contour image.
  • the k2' is a constant.
  • k' 2 k 2 , where k 2 is the spatial resolution in the width direction at a thickness of 0.
  • k 2 is the spatial resolution in the width direction at a thickness of 0.
  • the width value W' of the sawn timber to be tested expressed in the above contour image is the total number of pixel points of the measured cross-sectional contour of the sawn timber to be tested.
  • the beneficial effects of the present invention are: when the apparatus and method of the present invention measure, when the lens main plane of the area array industrial camera is parallel to the imaging plane, the incident laser line and the optical axis of the area array industrial camera are at an angle of 90°. Not only can the Gaussian imaging theorem be satisfied, but also multiple points on one section can be detected at the same time, so that the defect of the section can be quickly obtained.
  • the sawn timber advances with the conveyor at a constant speed, it can continuously detect any section of the sawn timber. The thickness information can thus quickly detect the defects of the entire surface of the sawn timber, and realize the detection of defects of the sawn timber.
  • Figure 1 is a schematic view showing the structure of a blunt edge sawing material
  • FIG. 2 is a schematic structural view of a blunt-edge sawn timber in a sawing state
  • Figure 3 - Figure 4 are graphs of warpage measurement
  • Figure 5 is a schematic view showing the system structure of the oblique sawing material defect detecting device according to the present invention.
  • Figure 6 is a schematic structural view of the oblique sawing material defect detecting device according to the present invention.
  • Figure 7 is a schematic view showing the process of detecting the blunt edge of the sawn timber
  • Figure 8 is a basic schematic diagram of the thickness of the sawn timber to be tested
  • Figure 9 is a schematic diagram of the detected width detection
  • Figure 10 is an image when the blunt edge sawing material is not placed
  • Figure 11 is an image when a blunt edge sawing material is placed
  • Figure 12 is a binarized image of Figure 11;
  • Figure 13 is a center line image of a blunt edge sawn timber
  • Figure 14 is a schematic view showing an incident angle of a laser light source
  • Figure 15 is a schematic diagram showing the curve of the thickness of the blunt edge sawing material
  • Figure 16 is a schematic view showing the curve of the width of the blunt edge sawing material
  • Figure 17 is a normal test Q-Q diagram of 45° error detection of a blunt edge sawing material
  • Figure 18 is a standard normal distribution quantizer for 45° error detection of a blunt edge sawn timber.
  • Figure 19 is a schematic view showing the flow of crack detection of the sawn timber to be tested.
  • Figure 20 is a crack detection image of the sawn timber to be tested
  • Figure 21 is a binarized image of Figure 21
  • Figure 22 is a center line image of a crack sawing material
  • Figure 23 is a schematic diagram showing the calibration curve of the crack sawing material thickness
  • Figure 24 is a schematic view showing the calibration curve of the crack sawing material width
  • Figure 25 is a normal test Q-Q diagram of 45° error detection of crack sawing material
  • Figure 26 is a standard normal distribution of the 45° error detection of cracked sawn timber.
  • Figure 27 is a schematic view showing the process of detecting the warpage of the sawn timber
  • Figure 28 is a warpage detection image of the sawn material to be tested
  • Figure 29 is a centerline image of a warped sawn timber
  • Figure 30 is a schematic view showing the curve of the thickness of the warped sawn timber
  • Figure 31 is a schematic diagram showing the curve of the width of the warped sawn timber.
  • the laser light source is from Prophotonix
  • the model number is D-660-010-0250-L01-S-90-S-S-2
  • the wavelength is 660 nm
  • the exit fan angle is 90 degrees
  • the power is 10 mW.
  • the area array industrial camera comes from PointGrey, model GS3-U3-23S6C-C
  • the camera resolution is 1920 ⁇ 1200
  • the maximum acquisition is 162 frames per second, USB3 interface.
  • the lens comes from: Nikon, model ML-U1614MP9, focal length 24mm.
  • FIG. 5 is a schematic structural view of a system for detecting an oblique sawing material defect according to the present invention
  • FIG. 6 is a schematic structural view of the oblique sawing material defect detecting device according to the present invention.
  • the oblique sawing material defect detecting device comprises a conveying table 1, a laser emitter 2, a bracket 3, an area array industrial camera 4, and the bracket 3 is bridged on the conveying table 1, along the conveying table.
  • the laser emitter 2 and the area array industrial camera 5 are sequentially disposed on the holder 3, and the light source emitted from the laser emitter 2 onto the surface of the sawn material 4 on the conveying table 1 is reflected by the sawn material 4 to be tested.
  • the area array industrial camera 5 receives, and the area array industrial camera 5 is simultaneously connected with a computer 7, the incident laser line emitted by the laser emitter 2 and the optical axis of the area array industrial camera 5 are at an angle of 90°, and the incident light is oblique. 0 ⁇ incident angle ⁇ 90 °.
  • Sawn timber defects are processing defects, specifically blunt edges.
  • Fig. 7 is a flow chart of a measuring method of the sawn timber to be tested.
  • the specific method for identifying the blunt edge of the sawn material 4 to be tested based on the above-mentioned identification device for the oblique type sawn timber includes the following steps:
  • the laser emitter 2 emits a laser light source to the sawn material 4 passing over the transfer table 1, and the laser light source emitted from the laser emitter 2 to the wooden board 4 to be tested is a fan-shaped laser light source formed on the surface of the sawn material 4 to be tested.
  • the light bar is a line of light, and the fan-shaped laser light source on the surface of the sawn material 4 is reflected and received by the array industrial camera 5, and the incident laser line emitted by the laser emitter 2 and the optical axis of the array industrial camera Incident angle of 90°, and the incident light is oblique, 0 ⁇ incident angle ⁇ 90°;
  • the area array industrial camera 5 converts the received large photoelectric signal into a digital signal and transmits it to the computer 7, and the data acquisition card in the computer 7 receives the digital signal, and is processed by the processing software of the computer to obtain the measurement of the sawn material 4 to be tested.
  • the contour image information of the cross section in the thickness and width directions is: the photoelectric signal is received by the area array industrial camera 5 and converted into a charge signal, and then converted into a digital image signal by external sampling amplification and analog-to-digital conversion circuit (specifically, Contour image information of thickness and width is binarized);
  • Figure 8 is a basic schematic diagram of the thickness of the sawn timber to be tested.
  • the laser emitter 2 emits a fan-shaped light source and is irradiated onto the horizontal transfer table 1.
  • the laser line is first imaged in the camera to be the clearest, and the entire test process no longer adjusts the distance during the imaging process.
  • the imaging process principle is shown in Fig. 7.
  • the laser emitter 2 emits a laser line to the surface of the sawn material 4 to be tested (in this embodiment, a blunt edge sawing material), an array of industrial cameras 5 and laser emission.
  • the device 2 is imaged at a certain angle.
  • the point O is the center of the lens lens, that is, the optical center
  • the point B is the point at which the laser light is irradiated onto the transfer table 1 when the sawn material is not placed, and the thickness value is detected as 0.
  • the image point on the imaging plane of the area array industrial camera 5 is B', and the line connecting BB' passes through the optical center perpendicular to the lens, that is, the optical axis.
  • the laser is irradiated at the point A of the surface of the sawn material to be tested, and the point on the imaging plane of the array industrial camera 5 is A', in the array industrial camera 5
  • the pixel value of the offset thickness is y' (unit: pixel)
  • k c is the pixel size (unit: mm/pixel) in the area array industrial camera 5, which is a camera constant.
  • the thickness of the sawn timber in the image is the offset thickness of the surface of the sawn timber product and the reference bottom surface of the industrial array camera.
  • ⁇ the angle between the optical axis of the area array industrial camera and the surface normal of the sawn timber to be tested
  • L represents the object distance imaged when the thickness is 0
  • L' is the image distance
  • L" is the object distance imaged when the thickness is not
  • f is the camera focal length.
  • x is the actual thickness value of the sawn timber to be tested
  • y is the offset of the surface of the sawn timber in the area array industrial camera from the reference bottom surface
  • y is expressed as the sawn timber in the thickness and width profile image.
  • the thickness, k is the spatial resolution in the thickness direction, and k is the linear relationship coefficient between the offset y of the surface of the sawn timber and the reference bottom surface of the sawn industrial camera and the thickness value x of the wooden board.
  • ⁇ is the incident angle of the laser emitter
  • ⁇ is the angle between the optical axis of the area array industrial camera and the normal to the surface of the sawn material
  • f is the camera focal length
  • L is the measured thickness when the thickness is 0.
  • the height of the table is determined, and L, f, and ⁇ are all fixed values, and k is also a fixed value.
  • Figure 9 is a schematic diagram of width detection.
  • W is the actual width value of the sawn timber to be tested
  • k2' is the spatial resolution of the width of the sawn timber to be tested
  • W' is the width of the sawn timber to be measured in the width profile image.
  • k2' is a constant.
  • Fig. 10 is an image when the sawn material to be tested is not placed. As shown in Fig. 10, there is only one laser line illuminated on the conveyor belt as a reference line for detecting the thickness of the blunt edge sawing material.
  • Figure 11 is an image when the sawn material is placed. As shown in Figure 11, there is only a black background and a very bright laser line.
  • Fig. 12 is an image after binarization of Fig. 11, and Fig. 13 is an image of a center line of a blunt-edge sawn timber, which can show the boundary point between the outer surface (retained sheet) and the narrow surface (blunt edge) of the sawn material to be tested.
  • the test pieces used in the experiment were 9 pieces of common blunt-edge sawn timber with different thicknesses.
  • the average width of the fir blunt-edge sawn timber was 180mm and the length was 800mm. Both had obvious narrow surface, and the thickness of the blunt edge of the fir was 20-60mm.
  • the interval is 5mm, a total of 9 pieces, the experiment number is 1-9, and the actual thickness of the test piece is measured by the detection system.
  • the thickness gauge is a thickness gauge with a specification of 5mm to 25mm, an interval of 0.5mm, 30mm to 60mm, and an interval of 5mm, for a total of 48 pieces.
  • the width-measured gauge block has a width of 100 mm, a thickness specification of 10 mm to 60 mm, and an interval of 5 mm, for a total of 11 pieces.
  • the thickness and width resolution curves of the experimental device are first calibrated, and then a mark line is drawn in the width direction of the blunt edge sawing material to be tested, and the laser line and the mark line are coincident during the test to obtain a laser Measurements. Then use the vernier caliper to measure the length of the blunt-edge sawing material mark line, and consider it to be accurate, that is, the actual value, and finally compare the error value between the actual value and the measured value to measure the accuracy and reliability of the detection system.
  • Fig. 14 is a schematic diagram of the incident angle of the laser light source.
  • the schematic diagram of the incident angle of the laser light source is shown in Fig. 14, keeping the laser emitter illuminated.
  • the normal line of the laser line on the transfer table is the same as the distance of the bracket of the industrial camera of the area array.
  • the measurement of 9 blunt-edge sawn timbers is performed, and the number of detections on each blunt edge sawing material is 56 lines.
  • x 1 the actual thickness of the blunt edge sawing material
  • y 1 the measuring block pixel value
  • k 1 the slope
  • Figure 15 is a schematic diagram of a thickness calibration curve.
  • the thickness direction calibration of the laser source at an incident angle of 45° has a very high linearity, and the square of the correlation coefficient is above 0.99, which is almost a straight line.
  • the variance of the model is 0.1 or less, which is very small, indicating the accuracy of the model. very good.
  • y 2 represents a unit pixel corresponding to the actual width ratio
  • x 2 is a blunt edge sawing material thickness unit (mm)
  • k 2 represents a regression slope
  • b 2 represents an intercept.
  • Regression model: y 2 k 2 * x 2 + b 2 .
  • the model parameters are shown in Table 4 above. At the same time, as shown in Fig.
  • the resolution in the width direction is basically a constant, which is only related to the optical parameters of the measurement system. It has nothing to do with the thickness and does not change with the thickness of the blunt-edge sawn timber to be tested, thus simplifying the calibration procedure. It can be applied to the blunt edge detection of various specifications, which greatly simplifies the workflow and reduces the work intensity.
  • Figure 17 is a normal test Q-Q diagram for 45° error detection
  • Figure 18 is a standard normal distribution quantile map for 45° error detection.
  • the measurement system is reliable, the relative error is below 1%, the error average is within 0.5mm, and the detection accuracy meets the wood processing requirements.
  • Sawn timber defects are growth defects, specifically cracks.
  • Fig. 19 is a schematic view showing the flow of crack detection of the sawn timber to be tested.
  • the method for detecting the surface crack of the sawn timber is basically the same as that of the first embodiment, and the difference is only in the step S3: according to the information of the change point of the thickness in the contour image information and the width information between the jump points.
  • Surface crack information of the sawn timber to be tested that is, according to FIG. 8, in the thickness measurement model, according to the formula (2), information on the sudden change in the thickness is obtained, specifically: when the cracked material has a crack defect, the thickness of the wood where the crack is located suddenly decreases. (A sudden change occurs). On the contour image, the curve of the thickness appears discontinuous and the change is severe, and it can be judged that there is a crack at the discontinuity.
  • the image when the cracked saw material is not placed is the same as that of the first embodiment, as shown in Fig. 10, in which only one laser line illuminated on the transfer table is used as a reference line for detecting the thickness of the sheet; Fig. 20 is the sawn material to be tested.
  • the crack detection image as the image after the sawn material, is a black background and a bright, broken laser line;
  • Figure 21 is the binarized image of Figure 20;
  • Figure 22 is the center of the crack sawing material
  • the line image shows the boundary point between the outer surface of the sheet and the crack.
  • the experimental device is mainly composed of a camera, a laser light source, a frame, a control device, a computer, and the like.
  • the laser source is from Prophotonix
  • the model is D-660-010-0250-L01-S-90-S-S-2
  • the wavelength is 660nm
  • the exit fan angle is 90 degrees
  • the power is 10mW.
  • the camera comes from PointGrey, model GS3-U3-23S6C-C
  • the camera resolution is 1920 ⁇ 1200
  • the maximum acquisition is 162 frames per second, USB3 interface.
  • the lens comes from: Nikon, model ML-U1614MP9, focal length 24mm.
  • the test piece used in the experiment was African Golden Silk Pomelo, also known as Teskin Lotus.
  • seven different thickness plates with crack defects were taken as experimental objects, numbered from 1 to 7. Since the experiment should consider the effect of different thickness on the width resolution of the thickness, the height of plate 1 to 7
  • the dimensions are designed to be seven levels, namely: 23.50mm, 30.29mm, 31.33mm, 33.96mm, 42.77mm, 47.75mm, 51.66mm.
  • the dimensions of the seven plates are roughly the same, 750mm long and 135mm wide, in each plate. A line with 55 crack defects was selected as the sample.
  • the thickness and width resolution curves of the experimental device firstly calibrate the thickness and width resolution curves of the experimental device, and then draw a mark line in the width direction of the saw material to be tested.
  • the laser line and the mark line are coincident to obtain the laser measurement value.
  • the vernier caliper uses the vernier caliper to measure the length of the plate mark line, and consider it to be accurate, that is, the actual value, and finally compare the error value between the actual value and the measured value to measure the accuracy and reliability of the detection system.
  • the incident angle of the laser light source is as shown in Fig. 14.
  • the normal and the array industrial camera that keep the laser emitter on the conveyor belt
  • the distance between the brackets is unchanged, and 7 plates are measured.
  • the number of inspections set on each plate is 55 lines.
  • Figure 23 is a schematic diagram showing the curve of the crack sawing material thickness. As shown in Fig. 23, the calibration of the thickness direction of the laser source at an incident angle of 45° has a very high linearity, and the square of the correlation coefficient is above 0.99, which is almost a straight line. The variance of the model is 0.1 or less, which is very small, indicating the accuracy of the model. very good.
  • y 2 represents the unit width corresponding to the actual width ratio
  • x 2 is the sheet thickness unit (mm)
  • k 2 represents the regression slope
  • b 2 represents the intercept.
  • Regression model: y 2 k 2 * x 2 + b 2 .
  • the model parameters are shown in Table 4 above.
  • Fig. 24 that when the angle between the optical axis and the incident laser is 90°, it can be seen that the resolution in the width direction is basically a constant, which is only related to the optical parameters of the measurement system. It has nothing to do with the thickness and does not change with the thickness of the solid wood board to be tested, thus simplifying the calibration procedure. It can be applied to the detection of burrs of various specifications, which greatly simplifies the workflow and reduces the work intensity.
  • the defect of the sawn timber to be tested is a sawing material drying defect, specifically warpage.
  • Fig. 27 is a schematic view showing the flow of warpage detection of the sawn timber to be tested.
  • the method for detecting the warpage of the sawn timber surface is basically the same as that of the first embodiment, and the difference is only in the S3 step: the maximum chord height h max and the sawn timber to be tested are calculated according to the contour image information. Measure the width ⁇ to obtain the warpage f ⁇ ,
  • FIG. 28 is a warpage detection image of the sawn material to be tested, a black background and a strip. A very bright laser line with warp curvature;
  • Figure 29 shows the centerline image of the warped sawn timber, showing the warpage of the outer surface of the sawn timber and the boundary point with the baseline.
  • the test pieces used in the experiment are 4 pieces of fir boards with different thicknesses of common specifications.
  • the average width of sawn timber is 180mm and the length is 800mm. Both have obvious blunt edges.
  • the thickness of sawn timber is 20 ⁇ 50mm, the interval is 10mm, totaling 4 pieces, the experiment number 1 to 4, the actual thickness of the test piece is measured by the inspection system.
  • Thickness gauges are used with thickness gauges ranging from 5mm to 25mm, 0.5mm intervals, 30mm to 60mm, and 5mm intervals, for a total of 48 pieces.
  • the width-measured gauge block has a width of 100 mm, a thickness specification of 10 mm to 60 mm, and an interval of 5 mm, for a total of 11 pieces.
  • the thickness and width resolution curves of the experimental device are first calibrated, and then a mark line is drawn in the width direction of the plate to be tested.
  • the laser line and the mark line are coincident to obtain a laser measurement value.
  • the vernier caliper uses the vernier caliper to measure the length of the plate mark line, and consider it to be accurate, that is, the actual value, and finally compare the error value between the actual value and the measured value to measure the accuracy and reliability of the detection system.
  • the incident angle of the laser light source is as shown in Figure 14 below, and the normal and the array industrial camera that keep the laser emitter on the conveyor belt.
  • the distance between the brackets is unchanged, and four plates are measured.
  • the number of inspections set on each plate is 56 lines.
  • Block thickness (mm) 45 degree measurement thickness (pixels) 5 31 5.5 34.5 6 36.5 6.5 38.5 7 43 7.5 46.5 8 47.5 8.5 53 9 56.5 9.5 57 10 61.5 10.5 63.5 11 67.5 11.5 70 12 73 12.5 76.5 13 79.5 13.5 82.5 14 85 14.5 89.5 15 92.5 15.5 94.5 16 97 16.5 100 17 103 17.5 105 18 109.5 18.5 113 19 116 19.5 119 20 121.5 20.5 125 twenty one 127.5 21.5 132 twenty two 136.5 22.5 138 twenty three 140 23.5 143.5 twenty four 144 24.5 148.5 25 151.5 30 183.5 35 242.5 40 305.5 45 368
  • x 1 the actual thickness of the warped sheet
  • y 1 the gauge block measurement pixel value
  • k 1 the slope
  • Figure 30 is a schematic diagram showing the curve of the thickness of the warped sawn timber.
  • the calibration of the thickness direction of the laser source with an incident angle of 45° has a very high linearity, and the square of the correlation coefficient is above 0.99, which is almost a straight line.
  • the variance of the model is 0.1 or less, which is very small, indicating the accuracy of the model. very good.
  • y 2 represents a unit pixel corresponding to the actual width ratio
  • x 2 is a warped sawn timber thickness unit (mm)
  • k 2 represents a regression slope
  • b 2 represents an intercept.
  • Regression model: y 2 k 2 * x 2 + b 2 .
  • the model parameters are shown in Table 16 above. At the same time, as shown in Fig.
  • the resolution in the width direction is basically a constant, which is only related to the optical parameters of the measurement system. It does not matter the thickness and does not change with the thickness of the sawn timber to be tested, thus simplifying the calibration procedure. It can be applied to the warpage detection of various specifications, which greatly simplifies the work flow and reduces the work intensity.
  • the measurement system is reliable, the warpage accuracy is 1%, and the maximum chord length accuracy is also 0.01, indicating that the present invention can realize on-line detection of woodboard warpage.
  • the warpage in the length direction passes the length of the wood board, and other methods remain unchanged.
  • the invention has the advantages and positive effects: compared with the manual detection method, the laser on-line detection has the advantages of high efficiency and high precision; at the same time, the triangulation method is adopted for measurement, the measurement precision is improved, and the industrial online detection is realized.

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Chemical & Material Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Biochemistry (AREA)
  • General Health & Medical Sciences (AREA)
  • Textile Engineering (AREA)
  • Immunology (AREA)
  • Pathology (AREA)
  • Length Measuring Devices By Optical Means (AREA)

Abstract

本发明公开了斜射式锯材缺陷检测装置和方法，包括激光发射器和面阵工业相机，激光发射器发射至锯材表面的光经反射后被面阵工业相机接收，入射激光线和面阵工业相机的光轴成90°夹角，入射光为斜射。本发明所述装置和方法测量精度高、使用方便，具有可同时检测一个截面上的多个点、自动连续检测的优势。

Description

一种斜射式锯材缺陷检测装置及检测方法 技术领域
本发明涉及一种检测装置及检测方法，尤其是涉及一种斜射式锯材缺陷的检测装置及其检测方法，属于测量技术领域。
背景技术
近年来,我国木材工业发展迅速，森林资源开采范围却在逐渐减少，森林资源面临着严重的短缺问题，木材的供需矛盾日益尖锐，这就对木材资源实现价值最大化提出了越来越高的要求。在木材资源缺乏的情况下,提高木材的利用率是木材加工与自动化领域中一个急需解决的问题。
而由于木材本身的特点和加工过程中的所处环境的影响，木材生长过程和后续加工处理过程中不可避免会产生缺陷。
如，生长缺陷中的裂纹现象，宽度尺寸较大的裂纹影响木材品质，影响外观，降低出材等级率。对实木板材表面裂纹进行检测并分类一直是木材检测领域的一项重要课题。
传统的检测往往由人工来完成，检测者用目视的方法判断缺陷类型，使用卡尺或其它量具来确定裂纹缺陷的尺寸，根据裂纹的宽度尺寸大小来确定板材的开裂程度。不仅工作量大，容易造成视觉疲劳，而且测量的过程受到检测人员主观因素的影响，检测的效率和质量难以保证，难以适应大规模工业自动化生产，特别是随着消费者对产品品质要求的不断提高，依靠人工检测测量越来越不能满足当今工业领域的要求。
如，加工缺陷中的钝棱现象：木材产品加工生产时会将原木锯切成一定厚度要求的板材，原木的截面可以近似看成圆形，而原木可以近似成一个圆柱形，一般锯切过程则是平行于圆柱轴线。经过锯切后的长板截面不是规则的矩形，而是两个侧面有不规则的曲边梯形。图1为钝棱板材的结构示意图，如图1所示：这种板材叫做钝棱板，钝棱锯材外材面分为上下两个宽材面和左右两个窄材面，由于原木本身不是一个规则的圆柱体、且有枝丫、弯曲等缺陷，使得钝棱锯材的窄面呈现不规则的形状，加工时需要将不规则窄材面去除。图2为锯切状态的钝棱板材的结构示意图，如图2所示：一般在后续的工段中往往要除去钝棱板的钝棱，如果去除窄材面不彻底就形成钝棱缺陷，影响锯材的等级和使用,如果去除过多将造成浪费。根据宽面到钝棱锯材髓心的远近程度分为内材面和外材面，距离髓心近的是内材面，远的是外材面。在许多锯材厂都需要有经验的工人，目视每一块锯材的外材面，凭借自己的经验进行齐边，工人的手工操作严重影响了加工精度和效率，并且造成钝棱锯材资源的浪费。
再如，干燥缺陷中的翘曲现象：木材的主要组成部分是纤维素、半纤维素和木素，对周围的水蒸气、水分有充分的亲和性，造就了木材天然的吸湿性能；木材是由细胞组成的生物体，内含许许多多的毛细管，具有可塑性。综上原因，木质板材或木制品会随着周围环境相对湿度的变化而产生干缩和湿胀现象。由于木材是一种非均质的各向异性材料，干缩与湿胀也具有各项异性，置于自然环境中的木质板材会产生翘曲，不同温湿度环境中的木质板材翘曲度不同。
图3-图4为翘曲度测量图。如图3和图4所示：根据国家标准GB/T 15036.2-2009实木地板中翘曲度的检验方法：将地板表面向上放置在水平实验台面上，把刀口直尺或钢板尺垂直紧靠地板两长边，用塞尺量取最大弦高h max，精确至0.01mm。最大弦高h max与地板实测宽度ω之比值为宽度方向翘曲度
Figure PCTCN2018094708-appb-000001
f ω以百分数表示，精确至0.01％。
传统的检测往往由人工来完成，检测者用塞尺或其它量具来确定木板翘曲的最大弦长，根据测量结果，手动计算木质板材翘曲度。工作量大，效率低，受主观测量影响，无法进行在线高效率的检测，难以投入工业量化检测。
针对以上问题，近年来迅速发展的以图像处理技术为基础的机器视觉技术，用摄像机对板材表面进行实时拍照，照片经数字化处理后送入计算机进行图象处理，通过参数计算对板材图片提取特征以检测表面缺陷信息，然后进行分类定等级。这种方法是用图像表征缺陷点的颜色信息来区分缺陷的，如果缺陷的图像颜色信息与木材图像信息差异较大时比较有效，但是木材往往会出现多种缺陷和本身颜色的变异，因而这种仅靠颜色信息来识别缺陷的方法会出现误判，造成一定的损失，且由于自身的局限，无法利用到板材缺陷后厚度和宽度方向上几何尺寸变化的根本特点，这是产生误差的主要原因。
目前，利用激光扫描法进行被测锯材缺陷的技术正在发展，激光扫描法利用被测锯材缺陷与其表面轮廓相关联，缺陷所在部位的厚度比正常锯材厚度薄的特征识别缺陷。具体步骤为激光位移传感器发射光源投向固定在工作台上的试件表面，而后由激光位移传感器接收试件表面的反射光，根据投射和反射光路径长短通过计算可得到测试点的厚度轮廓信息。激光位移传感器在水平方向可移动，从而可扫描并得到整个试件表面的厚度轮廓信息。通过专用处理软件把厚度轮廓转换为厚度轮廓图像，经过一系列图像处理和特征识别实现缺陷的定位和识别。这种方法可以准确测量出测量点处的缺陷信息，但是它也存在不足之处，比如检测速度慢，在实际检测时需要大量的测绘点采集才能得到木材缺陷轮廓信息，所以这种方法无法适应自动化生产的需要；同时该激光扫描法是一种激光时间差式，发出激光脉冲，测量脉冲从发出到返回的时间差，换算成探头到被测物体的距离，由于光的传播速度很快，准确测量出小位移量时对时间测量的精度要求非常高，否则测量的精度难以保证，而且传感器价格也较高。
发明内容
为了克服现有技术问题，本发明的目的在于提供一种安全可靠、测量精度高、使用方便的斜射式锯材缺陷检测装置。
一种斜射式锯材缺陷检测装置，包括传送台、激光发射器、支架、面阵工业相机，支架桥式架设在传送台上，沿传送台的传送方向，依次在支架上设置激光发射器和面阵工业相机，且所述的激光发射器发射到传送台上被测锯材表面的光源经被测锯材反射后被面阵工业相机接收，而面阵工业相机同时与一计算机连通，所述的激光发射器发射的入射激光线和面阵工业相机的光轴成90°夹角，，且所述的入射光为斜射，且，0＜入射角＜90°。
进一步，上述的锯材缺陷为锯材生长缺陷、锯材加工缺陷、锯材干燥缺陷中的任一种。
此外，上述的激光发射器发射到传送台上被测锯材表面的光源为扇形激光光源，、从而被测锯材表面形成光条，且所述的光条为一字形光条。
本发明还提供了一种基于上述的斜射式锯材缺陷检测装置的锯材缺陷的检测方法。
一种斜射式锯材缺陷检测方法，包括以下步骤：
S1：激光发射器向传送台上经过的被测锯材发射激光光源，经被测锯材表面反射后由面阵工业相机接 收，所述的激光发射器发射的入射激光线和面阵工业相机的光轴成90°夹角，且所述的入射光为斜射，且，0＜入射角＜90°；
S2：面阵工业相机将接收到的光电信号转换为数字信号后传输到计算机，计算机中的数据采集卡接收数字信号，并经过计算机的处理软件处理后得到被测锯材的测量截面在厚度和宽度方向上的轮廓图像信息，再根据得到的轮廓图像信息得到被测锯材的厚度值像素和宽度值像素，并将像素单位转化为长度单位；
S3：根据步骤S2得到的被测锯材的测量截面的轮廓信息以及像素单位和长度单位之间的换算关系，进一步得到锯材缺陷的位置信息，并根据锯材缺陷的位置信息进行相应的去除缺陷的操作。
更进一步，上述的数据处理软件中被测锯材的厚度的计算公式为：
x＝k·y      (I)
公式(I)中，x为被测锯材的检测厚度值；y为面阵工业相机中被测锯材表面与基准底面的偏移，y在轮廓图像中表现为被测锯材的厚度,k为厚度方向的空间分辨率，且所述的k为面阵工业相机中被测锯材表面与基准底面的偏移y与被测锯材的检测厚度值x的线性关系系数。
上述的k的数值公式为：
Figure PCTCN2018094708-appb-000002
在k的数值公式中，θ为激光发射器的入射角,β为面阵工业相机的光轴与被测锯材表面法线的夹角，f为相机焦距，L表示测量厚度为0时面阵工业相机成像的物距，θ+β＝90°，所以sin(θ+β)为固定值1，且当成像***固定，面阵工业相机的镜头位置固定且倾角确定，激光发射器距传送台高度确定，L、f、θ均为固定值，则k也为固定值。
而上述的被测锯材的宽度的计算公式为：
ω＝k2′W'      (Ⅱ)
公式(Ⅱ)中，所述的ω为被测锯材的实际宽度值，k2′为被测锯材宽度方向的空间分辨率，W'为轮廓图像中表现的被测锯材的宽度值，所述的k2′为常数。
且所述的k' 2＝k 2，式中：k 2为厚度为0处宽度方向的空间分辨率，当成像***固定，测量厚度为0时，成像的物距和像距均为固定值，则k 2为固定值。
而上述的轮廓图像中表现的被测锯材的宽度值W′为被测锯材的被测截面轮廓线的像素点总数。
本发明的有益效果为：本发明所述装置和方法在进行测量时，在面阵工业相机的透镜主平面与成像平面平行时，入射激光线和面阵工业相机的光轴成90°夹角，不仅能满足高斯成像定理，而且可同时检测一个截面上的多个点，从而快速得出该截面的缺陷情况，当锯材随传送台匀速前进时，可连续地检测锯材任一截面的厚度信息从而可快速地检测出锯材整个表面的缺陷情况，实现在线锯材缺陷的检测。克服了人工检测方法效率低，精确度低，自动化水平低的问题；机器视觉技术无法利用板材出现缺陷后厚度和宽度上几何尺寸变化，仅靠颜色信息来识别裂纹的方法精度较低，且不能实现在线检测；现有激光检测技术单点测量速度慢、无法在线工业化测量的缺点。
附图说明
图1为钝棱锯材的结构示意图；
图2为锯切状态的钝棱锯材的结构示意图；
图3-图4为翘曲度测量图；
图5为本发明所述的斜射式锯材缺陷检测装置的***结构示意图；
图6为本发明所述的斜射式锯材缺陷检测装置的结构示意图；
图7为被测锯材钝棱检测流程示意图；
图8为被测锯材厚度检测的基本原理图；
图9为被测宽度检测的原理图；
图10为未放入钝棱锯材时的图像；
图11为放入钝棱锯材时的图像；
图12为图11二值化后图像；
图13为钝棱锯材中心线图像；
图14为激光光源入射角示意图；
图15为钝棱锯材厚度标定曲线示意图；
图16为钝棱锯材宽度标定曲线示意图；
图17为钝棱锯材45°误差检测的正态检验Q-Q图；
图18为钝棱锯材45°误差检测的标准正态分布分位图。
图19为被测锯材裂纹检测流程示意图；
图20为被测锯材裂纹检测图像；
图21为图21的二值化图像；
图22为裂纹锯材中心线图像；
图23为裂纹锯材厚度标定曲线示意图；
图24为裂纹锯材宽度标定曲线示意图；
图25为裂纹锯材45°误差检测的正态检验Q-Q图；
图26为裂纹锯材45°误差检测的标准正态分布分位图。
图27为被测锯材翘曲检测流程示意图；
图28为被测锯材翘曲检测图像；
图29为翘曲锯材中心线图像；
图30为翘曲锯材厚度标定曲线示意图；
图31为翘曲锯材宽度标定曲线示意图。
图中主要附图标记含义为：
1、传送台        2、激光发射器  3、支架        4、被测锯材
5、面阵工业相机  6、PLC控制箱   7、数据采集卡  8、计算机。
具体实施方式
下面结合附图和具体实施例详细说明本发明。
在本实施方式中，激光光源出自：Prophotonix公司，型号是D-660-010-0250-L01-S-90-S-S-2，波 长是660nm，出光扇角是90度，功率10mW。面阵工业相机出自是PointGrey公司，型号GS3-U3-23S6C-C，相机分辨率是1920×1200，每秒钟最大采集162帧，USB3接口。镜头出自为：Nikon，型号ML-U1614MP9，焦距24mm。
图5为本发明所述的斜射式锯材缺陷检测装置的***结构示意图；图6为本发明所述的斜射式锯材缺陷检测装置的结构示意图。
如图5和图6所示：斜射式锯材缺陷检测装置，包括传送台1、激光发射器2、支架3、面阵工业相机4，支架3桥式架设在传送台1上，沿传送台1的传送方向，依次在支架3上设置激光发射器2和面阵工业相机5，且激光发射器2发射到传送台1上被测锯材4表面的光源经被测锯材4反射后被面阵工业相机5接收，而面阵工业相机5同时与一计算机7连通，激光发射器2发射的入射激光线和面阵工业相机5的光轴成90°夹角，且入射光为斜射，0＜入射角＜90°。
实施例1：
锯材缺陷为加工缺陷，具体为钝棱。
图7为被测锯材的测量方法流程图。
如图7所示：具体的基于上述的斜射式被测锯材的识别装置的被测锯材4钝棱的识别方法，包括以下步骤：
S1：激光发射器2向传送台1上经过的被测锯材4发射激光光源，且激光发射器2向被测木质板材4发射的激光光源为扇形激光光源，被测锯材4表面形成的光条为一字形光条，且被测锯材4表面的扇形激光光源经反射后由面阵工业相机5接收，所述的激光发射器2发射的入射激光线和面阵工业相机的光轴成90°夹角，且所述的入射光为斜射，0＜入射角＜90°；
S2：面阵工业相机5将接收大的光电信号转换为数字信号后传输到计算机7，计算机7中的数据采集卡接收数字信号，并经过计算机的处理软件处理后得到被测锯材4的测量截面在厚度和宽度方向上的轮廓图像信息，具体方法为：光电信号被面阵工业相机5接收并转化成电荷信号，再经过外部采样放大及模数转化电路转化成数字图像信号(具体为对厚度和宽度的轮廓图像信息进行二值化处理)；
其中厚度测量数学模型为：
图8为被测锯材厚度检测的基本原理图。
如图8所示：激光发射器2发射扇形光源，照射到水平的传送台1上，测量时先将激光线在相机中成像调至最清晰，整个测试过程不再调节成像过程中相距、物距等成像参数，成像过程原理如图7所示，激光发射器2发射激光线至被测锯材4(在本实施例中为钝棱锯材)的表面，面阵工业相机5与激光发射器2成一定角度来成像，图中O点是镜头透镜的中心，即光心，B点是没有放被测锯材时激光照射到传送台1上的点，此时检测厚度值为0，此时在面阵工业相机5成像平面上的像点是B'，BB'的连线通过光心垂直于透镜，即光轴。当放上一个厚度为x的被测锯材时，激光照射在被测锯材表面A点处，此时在面阵工业相机5成像平面上的点为A'，在面阵工业相机5中偏移厚度的像素值为y'(单位：像素)，k c为面阵工业相机5中像素大小(单位：mm/像素),为相机常数。令y＝y′·k c，代表相机中偏移厚度的像素值y'与相机常数k c的乘积，即图像中被测锯材4的厚度(单位：mm)。
x：检测厚度值
y：图像中被测锯材的厚度即面阵工业相机中被测锯材产品表面与基准底面的偏移厚度
θ：激光发射器的入射角
β：面阵工业相机的光轴与被测锯材的表面法线夹角
由图中几何关系可以知道：
x＝ABcosθ
y＝A'B'
L＝OB
L'＝OB'
x'＝AC
L”＝OC
其中L表示测量厚度为0时成像的物距，L'为像距，L”为测量厚度不为0时成像的物距，f为相机焦距。当相机位置固定时，L、f、θ为固定值，由放大率公式可得
Figure PCTCN2018094708-appb-000003
即：x＝k·y      (2)
公式(2)中，x为被测锯材的实际厚度值；y为面阵工业相机中被测锯材表面与基准底面的偏移，y在厚度和宽度轮廓图像中表现为被测锯材的厚度,k为厚度方向的空间分辨率，且k为面阵工业相机中被测锯材表面与基准底面的偏移y与木质板材检测厚度值x的线性关系系数，这个值大小与测试***中透镜焦距、未放被测锯材时物距、光轴与激光束的夹角共同决定的。
且上述的k的数值公式为：
Figure PCTCN2018094708-appb-000004
在k的数值公式中，θ为激光发射器的入射角,β为面阵工业相机的光轴与被测锯材表面法线的夹角，f为相机焦距，L表示测量厚度为0时面阵工业相机成像的物距，θ+β＝90°，所以sin(θ+β)为固定值1，且当成像***固定，面阵工业相机的镜头位置固定且倾角确定，激光发射器距传送台高度确定，L、f、θ均为固定值，则k也为固定值。
而宽度测量的数学模型则为：
图9为宽度检测的原理图。
如图9所示：由厚度方向测量模型推理中可以看出，当在水平传送台上放置一厚度x的被测锯材时，被测锯材表面的激光轮廓线垂直于光轴向透镜移动了一段距离至C点所在处的宽度方向上，如图9所示：
设k 2表示厚度为0处宽度方向的空间分辨率，k' 2为厚度为x处宽度方向的空间分辨率，由透镜公式可得：
Figure PCTCN2018094708-appb-000005
如公式(3)所示：θ+β＝90°，所以cos(θ+β)为0，所以所述的k' 2＝k 2，式中：k 2为厚度为0 处宽度方向的空间分辨率，当成像***固定，测量厚度为0时成像的物距和像距均为固定值，则k 2为固定值，而最终数据处理软件中被测锯材的宽度的计算公式为：
W＝k2′W'      (4)
公式(4)中，所述的W为被测锯材的实际宽度值，k2′为被测锯材宽度方向的空间分辨率，W'为宽度轮廓图像中表现的被测锯材的宽度值，所述的k2′为常数。
图10为未放入被测锯材时的图像。如图10所示：图中只有一条传送带上被照亮的激光线，作为检测钝棱锯材厚度的参照线。
图11为放入被测锯材时的图像。如图11所示：只有黑色的背景和一条很亮的激光线。
图12为图11二值化后图像，图13为钝棱锯材中心线图像，可显示出被测锯材的外材面(留用的板材)与窄材面(钝棱)的边界点。
S3：根据被测锯材的厚度和宽度信息确定锯材的正常板面与钝棱的边界点，每隔一定长度检测出一个边界点，以同一侧测出的边界点中最内侧的边界点为锯切点，沿平行于正常板面中心线的锯路去除钝棱。
以上实施例的具体标定过程为：
(1)实验材料：
实验所用的试件为9块常见不同规格厚度的杉木钝棱锯材，杉木钝棱锯材的平均宽度180mm,长度800mm,均有明显的窄材面，杉木钝棱厚度规格为20～60mm,间隔5mm,共计9块，实验编号1～9，试件实际厚度由检测***测得。厚度标定使用厚度规，规格为5mm～25mm，间隔0.5mm，30mm～60mm，间隔5mm，共计48块。宽度标定的量块宽为100mm，厚度规格为10mm～60mm，间隔5mm，共计11块。
(2)实验方法：
根据上述的检测实验模型，首先标定实验装置的厚度和宽度分辨率曲线，然后在待测规格钝棱锯材宽度方向上画出一条记号线，测试时让激光线与记号线重合，得出激光测量值。再用游标卡尺测量钝棱锯材记号线长度，并认为是准确的，即为实际值，最后比较实际值与测量值之间的误差值，来衡量检测***的精度和可靠性。面阵工业相机的倾斜角度为45°，激光光源入射角θ＝45°情况下进行实验，图14为激光光源入射角示意图，激光光源入射角示意图如图14所示，保持激光发射器照射在传送台上激光线的法线与面阵工业相机所在支架的距离不变，进行9块钝棱锯材的测量，每一块钝棱锯材上设定检测数目为56条线。
(3)实验数据：
表1 厚度测量数据
Figure PCTCN2018094708-appb-000006
表2 厚度模型参数表
Figure PCTCN2018094708-appb-000007
由表1和表2可知：x 1代表钝棱锯材实际厚度，y 1代表量块测量像素值，k 1代表斜率，b 1代表截距厚度回归方程为:x 1＝k 1*y 1+b 1。
图15为厚度标定曲线示意图。
如图15所示：在激光光源入射角为45°下的厚度方向标定具有非常高的线性度，相关系数平方都在0.99以上，几乎是一条直线，模型方差0.1以下，非常小，说明模型精度非常好。
表3 45度宽度标定值
Figure PCTCN2018094708-appb-000008
表4:宽度模型参数表
倾角 k2 b2 R 2相关系数平方 σ 2模型方差
45度 0.00000036916 0.19544 0.071602 5.38e -08
如表3和表4所示：y 2代表单位像素对应实际宽度比值，x 2是钝棱锯材厚度单位(mm),k 2代表回归斜率，b 2代表截距。回归模型：y 2＝k 2*x 2+b 2。θ＝45°角度的相关系数在0.071左右，即量块实际厚度值与宽度分辨率几乎无线性关系，同时模型方差基本为0，几乎是一条水平直线。模型参数见上表4所示，同时如图16可知，当光轴与入射激光夹角为90°时，可以看出宽度方向的分辨率基本为一常数，只于测量***的光学参数有关，而与厚度并无关系，不随被测钝棱锯材的厚度变化而变化，因而可以简化标定程序。可以适用于各种规格不同的板材钝棱检测，极大地简化了工作流程，降低了工作强度。
表5：45°相对误差均值
Figure PCTCN2018094708-appb-000009
表6：误差样本均值与方差
  均值 方差
45° 0.2191 0.296
图17为45°误差检测的正态检验Q-Q图；图18为45°误差检测的标准正态分布分位图。
如表5、6和图17和图18所示：测量***可靠，相对误差在1％以下，误差均值在0.5mm以内，检测精度满足木材加工需求。
实施例2
锯材缺陷为生长缺陷，具体为裂纹。
图19为被测锯材裂纹检测流程示意图。
如图19所示：被测锯材表面裂纹的检测方法与实施例1基本相同，区别仅在于：S3步骤中：根据轮廓图像信息中厚度出现突变点的信息和突变点之间的宽度信息得到被测锯材的表面裂纹信息。即根据图8所示，在厚度测量模型中，根据公式(2)，得到厚度出现突变点的信息，具体为：当被测锯材存在裂纹缺陷时，裂纹所在处的木材厚度会突然减小(出现突变点)，在轮廓图像上，厚度的曲线表现为不连续，变化剧烈，即可判断在间断处存在裂纹。
未放入裂纹锯材时的图像与实施例1相同，如图10所示，图中只有一条传送台上被照亮的激光线，作为检测板材厚度的参照线；图20为被测锯材裂纹检测图像，作为放入锯材后的图像，其为只有黑色的背景和一条很亮的、有断点的激光线；图21为图20的二值化图像；图22为裂纹锯材中心线图像，可显示出板材外材面与裂纹的边界点。
本实施例中，厚度标度与宽度标定原理与实施例1相同。
本实施例的具体标定过程为：
(1)实验装置
实验装置主要由相机、激光光源、机架、控制装置、计算机等构成。激光光源出自：Prophotonix公司，型号是D-660-010-0250-L01-S-90-S-S-2，波长是660nm，出光扇角是90度，功率10mW。相机出自是PointGrey公司，型号GS3-U3-23S6C-C，相机分辨率是1920×1200，每秒钟最大采集162帧，USB3接口。镜头出自为：Nikon，型号ML-U1614MP9，焦距24mm。
(2)实验材料
实验所用的试件为非洲金丝柚，又名特斯金莲木。实验中，取七块材面存在裂纹缺陷的不同厚度的板材作为实验对象，编号为1至7号，由于实验需考虑不同厚度对该厚度所在的宽度分辨率的影响，1至7号板材高度尺寸设计为七个水平，分别为：23.50mm、30.29mm、31.33mm、33.96mm、42.77mm、47.75mm、51.66mm，7块板材的幅面尺寸大致相同，长750mm、宽135mm，在每块板材上选取55条有裂纹缺陷的直线作为样本。
(3)实验方法
根据上述的检测实验模型，首先标定实验装置的厚度和宽度分辨率曲线，然后在待测规格锯材宽度方向上画出一条记号线，测试时让激光线与记号线重合，得出激光测量值。再用游标卡尺测量板材记号线长度，并认为是准确的，即为实际值，最后比较实际值与测量值之间的误差值，来衡量检测***的精度和可靠性。相机的倾斜角度为45°，激光光源入射角θ＝45°情况下进行实验，激光光源入射角示意图如图14所示，保持激光发射器照射在传送带上激光线的法线与面阵工业相机所在支架的距离不变，进行7块板材的测量，每一块板材上设定检测数目为55条线。
(4)实验数据
表7 厚度标定测量数据
Figure PCTCN2018094708-appb-000010
表8、厚度模型参数表
Figure PCTCN2018094708-appb-000011
由表7和表8所示：x 1代表板材实际厚度，y 1代表量块测量像素值，k 1代表斜率，b 1代表截距厚度回归方程为:x 1＝k 1*y 1+b 1。
图23为裂纹锯材厚度标定曲线示意图。如图23所示，在激光光源入射角为45°下的厚度方向标定具有非常高的线性度，相关系数平方都在0.99以上，几乎是一条直线，模型方差0.1以下，非常小，说明模型精度非常好。
表9 45度宽度标定值
Figure PCTCN2018094708-appb-000012
表10、宽度模型参数表
倾角 k2 b2 R 2相关系数平方 σ 2模型方差
45度 0.00000036916 0.19544 0.071602 5.38 e-08
如表9和表10所示：
y 2代表单位像素对应实际宽度比值，x 2是板材厚度单位(mm),k 2代表回归斜率，b 2代表截距。回归模型：y 2＝k 2*x 2+b 2。θ＝45°角度的相关系数在0.071左右，即量块实际厚度值与宽度分辨率几乎无线性关系，同时模型方差基本为0，几乎是一条水平直线。模型参数见上表4所示，同时从图24可知：当光轴与入射激光夹角为90°时，可以看出宽度方向的分辨率基本为一常数，只于测量***的光学参数有关，而与厚度并无关系，不随被测实木板材的厚度变化而变化，因而可以简化标定程序。可以适用于各种规格不同的板材毛边检测，极大地简化了工作流程，降低了工作强度。
表11、45°相对误差均值
Figure PCTCN2018094708-appb-000013
表12、误差样本均值与方差
  均值 方差
45° 0.1821 0.1321
由表11、表12和图25和图26可知，测量***可靠，相对误差在1％以下，误差均值在0.5mm以内，检测精度满足木材加工需求。
实施例3：
被测锯材的缺陷为锯材干燥缺陷，具体为翘曲。
图27为被测锯材翘曲检测流程示意图。
如图27所示：被测锯材表面翘曲度的检测方法与实施例1基本相同，区别仅在于:S3步骤中：根据轮廓图像信息经计算得到最大弦高h max和被测锯材的实测宽度ω，得到翘曲度f ω，
Figure PCTCN2018094708-appb-000014
如实施例1中图10所示，图中只有一条传送带上被照亮的激光线，作为被测锯材厚度的参照线；图28为被测锯材翘曲检测图像，黑色的背景和一条很亮的具翘曲弧度的激光线；图29为翘曲锯材中心线图像，可显示出锯材外材面的翘曲弧度以及与基准线的边界点。
本实施例中，厚度标度与宽度标定原理与实施例1相同。
本实施例的具体标定过程为：
(1)实验材料：
实验所用的试件为4块常见不同规格厚度的杉木板材，锯材平均宽度180mm,长度800mm,均有明显的钝棱，锯材厚度规格为20～50mm,间隔10mm,共计4块，实验编号1～4，试件实际厚度由检测***测得。厚 度标定使用厚度规，规格为5mm～25mm，间隔0.5mm，30mm～60mm，间隔5mm，共计48块。宽度标定的量块宽为100mm，厚度规格为10mm～60mm，间隔5mm，共计11块。
(2)实验方法：
根据上述的检测实验模型，首先标定实验装置的厚度和宽度分辨率曲线，然后在待测规格板材宽度方向上画出一条记号线，测试时让激光线与记号线重合，得出激光测量值。再用游标卡尺测量板材记号线长度，并认为是准确的，即为实际值，最后比较实际值与测量值之间的误差值，来衡量检测***的精度和可靠性。相机的倾斜角度为45°，激光光源入射角θ＝45°情况下进行实验，激光光源入射角示意图如下图14所示，保持激光发射器照射在传送带上激光线的法线与面阵工业相机所在支架的距离不变，进行4块板材的测量，每一块板材上设定检测数目为56条线。
(3)实验数据：
表13 厚度测量数据
量块厚度(mm) 45度测量厚度(像素)
5 31
5.5 34.5
6 36.5
6.5 38.5
7 43
7.5 46.5
8 47.5
8.5 53
9 56.5
9.5 57
10 61.5
10.5 63.5
11 67.5
11.5 70
12 73
12.5 76.5
13 79.5
13.5 82.5
14 85
14.5 89.5
15 92.5
15.5 94.5
16 97
16.5 100
17 103
17.5 105
18 109.5
18.5 113
19 116
19.5 119
20 121.5
20.5 125
21 127.5
21.5 132
22 136.5
22.5 138
23 140
23.5 143.5
24 144
24.5 148.5
25 151.5
30 183.5
35 242.5
40 305.5
45 368
表14、厚度模型参数表
k1 b1 R 2相关系数平方 σ 2模型方差
0.16373 0.022099 0.9998 0.024153
由表13和表14可知：x 1代表翘曲板材的实际厚度，y 1代表量块测量像素值，k 1代表斜率，b 1代表截距厚度回归方程为:x 1＝k 1*y 1+b 1。
图30为翘曲锯材厚度标定曲线示意图。
如图30所示：在激光光源入射角为45°下的厚度方向标定具有非常高的线性度，相关系数平方都在0.99以上，几乎是一条直线，模型方差0.1以下，非常小，说明模型精度非常好。
表15 45度宽度标定值
Figure PCTCN2018094708-appb-000015
表16、宽度模型参数表
倾角 k2 b2 R 2相关系数平方 σ 2模型方差
45度 3.6916 e-07 0.19544 0.071602 5.38 e-08
如表15和表16所示：y 2代表单位像素对应实际宽度比值，x 2是翘曲锯材厚度单位(mm),k 2代表回归斜率，b 2代表截距。回归模型：y 2＝k 2*x 2+b 2。θ＝45°角度的相关系数在0.071左右，即量块实际厚度值与宽度分辨率几乎无线性关系，同时模型方差基本为0，几乎是一条水平直线。模型参数见上表16所示，同时如图31可知，当光轴与入射激光夹角为90°时，可以看出宽度方向的分辨率基本为一常数，只于测量***的光学参数有关，而与厚度并无关系，不随被测锯材的厚度变化而变化，因而可以简化标定程序。可以适用于各种规格不同的板材翘曲度检测，极大地简化了工作流程，降低了工作强度。
表17 45°翘曲度实测数据
Figure PCTCN2018094708-appb-000016
由表17可知，测量***可靠，翘曲度精度达到1％，最大弦长精度也在0.01，表示本发明可以实现木质板材翘曲度的在线检测。长度方向翘曲度则将木质板材长度方向通过，其他方法不变。本发明具有的优点和积极效果是：与人工检测方法相比，激光在线检测具有效率高，精度高等优点；同时采用三角测距方法进行测量，测量精度提高，实现工业在线检测。
本发明按照上述实施例进行了说明应当理解，上述实施例不以任何形式限定本发明，凡采用等同替换或等效变换方式所获得的技术方案，均落在本发明的保护范围之内。

Claims (9)

  1. 一种斜射式锯材缺陷检测装置，包括传送台、激光发射器、支架、面阵工业相机，支架桥式架设在传送台上，沿传送台的传送方向，依次在支架上设置激光发射器和面阵工业相机，且所述的激光发射器发射到传送台上被测锯材表面的光源经被测锯材反射后被面阵工业相机接收，而面阵工业相机同时与一计算机连通，其特征在于，所述的激光发射器发射的入射激光线和面阵工业相机的光轴成90°夹角，，且所述的入射光为斜射，且，0＜入射角＜90°。
  2. 根据权利要求1所述的一种斜射式锯材缺陷检测装置，其特征在于，所述的锯材缺陷为锯材生长缺陷、锯材加工缺陷、锯材干燥缺陷中的任一种。
  3. 根据权利要求1所述的一种斜射式锯材缺陷检测装置，其特征在于，所述的激光发射器发射到传送台上被测锯材表面的光源为扇形激光光源，、从而被测锯材表面形成光条，且所述的光条为一字形光条。
  4. 一种基于权利要求1-3任一项权利要求所述的斜射式锯材缺陷检测装置的斜射式锯材缺陷检测方法，其特征在于，包括以下步骤：
    S1：激光发射器向传送台上经过的被测锯材发射激光光源，经被测锯材表面反射后由面阵工业相机接收，所述的激光发射器发射的入射激光线和面阵工业相机的光轴成90°夹角，且所述的入射光为斜射，且，0＜入射角＜90°；
    S2：面阵工业相机将接收到的光电信号转换为数字信号后传输到计算机，计算机中的数据采集卡接收数字信号，并经过计算机的处理软件处理后得到被测锯材的测量截面在厚度和宽度方向上的轮廓图像信息，再根据得到的轮廓图像信息得到被测锯材的厚度值像素和宽度值像素，并将像素单位转化为长度单位；
    S3：根据步骤S2得到的被测锯材的测量截面的轮廓信息以及像素单位和长度单位之间的换算关系，进一步得到锯材缺陷的位置信息，并根据锯材缺陷的位置信息进行相应的去除缺陷的操作。
  5. 根据权利要求4所述的一种斜射式锯材缺陷检测方法，其特征在于，所述的数据处理软件中被测锯材的厚度的计算公式为：x＝k·y       (I)
    公式(I)中，x为被测锯材的检测厚度值；y为面阵工业相机中被测锯材表面与基准底面的偏移，y在轮廓图像中表现为被测锯材的厚度,k为厚度方向的空间分辨率，且所述的k为面阵工业相机中被测锯材表面与基准底面的偏移y与被测锯材的检测厚度值x的线性关系系数。
  6. 根据权利要求5所述的一种斜射式锯材缺陷检测方法，其特征在于，所述的k的数值公式为：
    Figure PCTCN2018094708-appb-100001
    在k的数值公式中，θ为激光发射器的入射角,β为面阵工业相机的光轴与被测锯材表面法线的夹角，f为相机焦距，L表示测量厚度为0时面阵工业相机成像的物距，θ+β＝90°，所以sin(θ+β)为固定值1，且当成像***固定，面阵工业相机的镜头位置固定且倾角确定，激光发射器距传送台高度确定，L、f、θ均为固定值，则k也为固定值。
  7. 根据权利要求5所述的一种斜射式锯材缺陷检测方法，其特征在于，所述的被测锯材的宽度的计算公式为：
    ω＝k 2′W'   (Ⅱ)
    公式(Ⅱ)中，所述的ω为被测锯材的实际宽度值，k 2′为被测锯材宽度方向的空间分辨率，W'为轮廓图像中表现的被测锯材的宽度值，所述的k 2′为常数。
  8. 根据权利要求7所述的一种斜射式锯材缺陷检测方法，其特征在于，所述的k' 2＝k 2，式中：k 2为厚度为0处宽度方向的空间分辨率，当成像***固定，测量厚度为0时，成像的物距和像距均为固定值，则k 2为固定值。
  9. 根据权利要求7所述的一种斜射式锯材缺陷检测方法，其特征在于，所述的轮廓图像中表现的被测锯材的宽度值W′为被测锯材的被测截面轮廓线的像素点总数。
PCT/CN2018/094708 2017-07-06 2018-07-05 一种斜射式锯材缺陷检测装置及检测方法 WO2019007405A1 (zh)

Applications Claiming Priority (6)

Application Number Priority Date Filing Date Title
CN201710545226.3A CN107388973A (zh) 2017-07-06 2017-07-06 一种斜射式木质板材毛边的识别装置和识别方法
CN201710545221.0 2017-07-06
CN201710545226.3 2017-07-06
CN201710545229.7 2017-07-06
CN201710545221.0A CN107388962A (zh) 2017-07-06 2017-07-06 一种木质板材翘曲的在线检测装置和检测方法
CN201710545229.7A CN107561091A (zh) 2017-07-06 2017-07-06 一种斜射式实木板材表面裂纹的检测***和检测方法

Publications (1)

Publication Number Publication Date
WO2019007405A1 true WO2019007405A1 (zh) 2019-01-10

Family

ID=64949748

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/094708 WO2019007405A1 (zh) 2017-07-06 2018-07-05 一种斜射式锯材缺陷检测装置及检测方法

Country Status (1)

Country Link
WO (1) WO2019007405A1 (zh)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1040265A (zh) * 1988-08-10 1990-03-07 北京科技大学 激光扫描测粗糙度的方法及测量仪
CN103322948A (zh) * 2013-07-04 2013-09-25 上海宝锋工程技术有限公司 板坯表面凹凸点检测方法、检测装置与处理方法
CN106018431A (zh) * 2016-05-23 2016-10-12 南京林业大学 一种实木板材表面裂纹的检测***和检测方法
CN107388962A (zh) * 2017-07-06 2017-11-24 南京林业大学 一种木质板材翘曲的在线检测装置和检测方法
CN107388973A (zh) * 2017-07-06 2017-11-24 南京林业大学 一种斜射式木质板材毛边的识别装置和识别方法
CN107561091A (zh) * 2017-07-06 2018-01-09 南京林业大学 一种斜射式实木板材表面裂纹的检测***和检测方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1040265A (zh) * 1988-08-10 1990-03-07 北京科技大学 激光扫描测粗糙度的方法及测量仪
CN103322948A (zh) * 2013-07-04 2013-09-25 上海宝锋工程技术有限公司 板坯表面凹凸点检测方法、检测装置与处理方法
CN106018431A (zh) * 2016-05-23 2016-10-12 南京林业大学 一种实木板材表面裂纹的检测***和检测方法
CN107388962A (zh) * 2017-07-06 2017-11-24 南京林业大学 一种木质板材翘曲的在线检测装置和检测方法
CN107388973A (zh) * 2017-07-06 2017-11-24 南京林业大学 一种斜射式木质板材毛边的识别装置和识别方法
CN107561091A (zh) * 2017-07-06 2018-01-09 南京林业大学 一种斜射式实木板材表面裂纹的检测***和检测方法

Similar Documents

Publication Publication Date Title
CN107561091A (zh) 一种斜射式实木板材表面裂纹的检测***和检测方法
CN107388962A (zh) 一种木质板材翘曲的在线检测装置和检测方法
AU2013314863B2 (en) Detection system based on modulation of line structured laser image of glass
CN106018431A (zh) 一种实木板材表面裂纹的检测***和检测方法
CN105806247A (zh) 一种木质板材翘曲的在线检测装置和检测方法
CN102095658B (zh) 片烟质量的检测方法与***
KR101019099B1 (ko) 비젼 카메라를 이용한 강재의 길이 및 너비 측정장치와,이를 이용한 강재의 길이 및 너비 측정방법
CN109975319B (zh) 一种平面光学元件表面质量快速检测装置及其方法
CN105783746A (zh) 一种木质产品厚度的检测***及其检测方法
CN106123789A (zh) 一种木质板材毛边的识别装置和识别方法
CN106767452A (zh) 一种木质产品宽度检测装置及其检测方法
Antikainen et al. Simultaneous measurement of lathe check depth and the grain angle of birch (Betula pendula Roth) veneers using laser trans-illumination imaging
CN115713524A (zh) 一种基于点云数据的木材破损类缺陷检测方法
CN103759675A (zh) 一种用于光学元件非球面微结构的同步检测方法
CN105403578A (zh) 一种基于x射线计算机断层扫描技术的竹材含水率快速检测方法
CN112345540B (zh) 一种基于圆形结构光检测内孔过盈配合表面拆卸损伤的检测方法
CN101464255B (zh) 定量性应力应变偏光测定机
WO2019007405A1 (zh) 一种斜射式锯材缺陷检测装置及检测方法
CN201548206U (zh) 玻璃厚度测量仪
CN109115747B (zh) 基于拉曼光谱和oct测定玻璃材料性质的***与方法
CN204154609U (zh) 水松纸透气度检测仪
CN107388973A (zh) 一种斜射式木质板材毛边的识别装置和识别方法
CN104865271A (zh) 一种区分透明薄板上下表面图像的检测方法
Xu et al. An image-based laser triangulation width model applied in Sawn Lumber’s external face measurement
CN103926257A (zh) 一种基于太赫兹时域光谱仪的镜片缺陷的检测方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18827773

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18827773

Country of ref document: EP

Kind code of ref document: A1