Mirror-like object defect detection device and detection method thereof

Abstract

The invention relates to a defect detection device for a mirror-like object and a detection method thereof. The device comprises an optical detection system, an industrial computer, a trigger controller, a motion controller, and a three-axis servo module. The industrial computer is connected with the motion controller. The three-axis servo module can drive the optical detection system to scan the detected object under the action of the industrial computer and the motion controller. The trigger controller can be connected with the three-axis servo module and the optical detection system respectively. The optical detection system is connected with the industrial computer. The optical detection system is used to collect and analyze surface data of the mirror-like object and comprises a linear array camera, an LED stripe light source, a plurality of diffuse reflection light sources, a Y-direction distance measurement and color difference detection component, and an X-direction distance measurement component. The defect detection device for a mirror-like object and the detection method thereof have high collection efficiency, high imaging quality, compact system structure, and low cost.

Description

Technical Field

[0001] The present invention relates to the field of defect detection and machine vision technology, and in particular to a defect detection device for a mirror-like object and a detection method thereof. Background Art

[0002] With the rapid development of materials science and technology, mirror and mirror-like materials are widely used on the surfaces of various products, such as automobile bodies, mobile phone casings, industrial bearing rollers, etc.

[0003] Traditional manual visual inspection is increasingly unable to meet the measurement and inspection requirements of the process due to its low inspection efficiency, lack of objective inspection standards, and susceptibility to subjective factors such as manual distraction and fatigue. The demand for mirror-like defect detection devices in various industries is increasing. This automated device can not only strictly control product quality, but also trace product defects in a timely manner, providing data support for process quality improvement.

[0004] There are many types of defects in mirror-like objects, and different manufacturers have different definitions of defects. From the optical imaging form of defects, they can be classified into: chromatic aberration defects, dirt defects, texture defects, scratch and bump defects, and concave and convex defects. A single 2D imaging method and detection method is difficult to deal with all types of defects. To detect all defects at the same time, it is often necessary to combine 2D imaging and 3D imaging methods. In 3D imaging, laser triangulation and fringe projection are height reconstructions. Three-dimensional reconstruction equipment based on the fringe projection principle is mainly used for diffuse reflection objects. Laser triangulation can be applied to the height measurement of mirror-like objects, but it is difficult to detect micron-level defects. In 3D imaging, photometric stereo and fringe reflection (phase measurement deflectometry) are gradient reconstructions. Photometric stereo based on the Lambertian illumination model has high accuracy in gradient reconstruction of diffuse reflection surfaces, but it is difficult to directly apply to mirror objects. Phase measurement deflectometry has high accuracy in gradient reconstruction of mirror objects, which can reach sub-micron level in principle. Monocular phase measurement deflectometry uses unwrapping methods such as Gray code or three-frequency extrapolation, which requires the acquisition of more pictures and has low hardware execution efficiency. Binocular phase measurement deflectometry introduces a second camera to solve the problem of sag uncertainty, but has low calculation and storage efficiency in software. Summary of the invention

[0005] In view of the above problems, in order to detect all types of defects on mirror surfaces and mirror-like surfaces, while ensuring micron-level detection accuracy, the present invention takes into account factors such as detection efficiency, system cost and compact structure. The present invention proposes a defect detection device and a detection method for mirror-like objects with high acquisition efficiency, high imaging quality, compact system structure and low cost.

[0006] The present invention is achieved through the following technical scheme: a defect detection device for mirror-like objects, comprising an optical detection system, an industrial computer, a trigger controller, a motion controller and a three-axis servo module, wherein the industrial computer is connected to the motion controller, and the three-axis servo module can drive the optical detection system to scan the object to be detected under the action of the industrial computer and the motion controller, the trigger controller can be connected to the three-axis servo module and the optical detection system respectively, and the optical detection system is connected to the industrial computer.

[0007] The optical detection system is used to collect and analyze the surface data of mirror-like objects, and includes a linear array camera, an LED stripe light source, a plurality of diffuse reflection light sources, a Y-direction distance measurement and color difference detection component, and an X-direction distance measurement component.

[0008] The linear array camera and the LED stripe light source form a phase measurement deflectometry system for calculating gloss, wrapping phase, gradient map, curvature map, etc. The linear array camera and the LED stripe light source are symmetrical about the surface normal of the detected object.

[0009] Preferably, the angle between the line array camera and the LED stripe light source is 10° to 60°.

[0010] The linear array camera and the diffuse reflection light source form a photometric stereo system for calculating albedo, surface normal vector, and assisting deflectometry unwrapping. The diffuse reflection light source is evenly distributed around the surface normal of the detected object.

[0011] Preferably, the angle between the diffuse reflection light source and the surface normal of the detected object is 60° to 80°.

[0012] Furthermore, the diffuse reflection light source includes a cooling fan, a heat conducting material, a patch LED, a focusing rod and a diffusion film, and the light of the patch LED is emitted obliquely from the light source surface at an angle of 40° to 50°.

[0013] Preferably, there are four diffuse reflection light sources. Depending on actual conditions, more or fewer diffuse reflection light sources may be designed, but at least three.

[0014] The Y-direction distance measurement and color difference detection component includes two Y-direction line lasers, a color difference detection light source and a Y-direction area array camera.

[0015] In the Y-direction ranging and chromatic aberration detection component, two Y-direction line lasers and a Y-direction area array camera constitute a laser triangulation ranging system, which is used to assist the unwrapping of deflectometry. The two Y-direction line lasers are coplanar with the line array camera and are distributed on both sides of the line array camera. The Y-direction area array camera and the Y-direction line laser are respectively arranged on both sides of the surface normal of the object to be detected.

[0016] Preferably, the angle between the Y-axis area array camera and the surface normal of the detected object is 40° to 60°.

[0017] In the Y-direction distance measurement and color difference detection component, the Y-direction area array camera and the color difference detection light source are symmetrical about the surface normal of the detected object and are used to detect color difference.

[0018] The X-direction distance measurement component includes an X-direction line laser and an X-direction array camera, which together form a laser triangulation distance measurement system, which is used to guide the movement of the three-axis servo module so that the optical detection system can adapt to the height change of the detection surface in the Z direction and the rotation change in the RY direction. The X-direction line laser is perpendicular to the YOZ plane of the coordinate system, and the X-direction array camera and the X-direction line laser are respectively arranged on both sides of the normal line of the surface of the detected object.

[0019] Preferably, the angle between the X-axis area array camera and the surface normal of the detected object is 40° to 60°.

[0020] The main working principle of the device is that the industrial computer communicates with the motion controller to make the three-axis servo module move in the X direction, the motion controller communicates with the trigger controller, the trigger controller communicates with the camera and light source in the optical detection system, and the camera in the optical detection system transmits image information to the industrial computer.

[0021] The present invention also includes a method for detecting defects of a mirror-like object, using the above-mentioned device for detecting defects of a mirror-like object, comprising the following steps:

[0022] 1) The industrial computer communicates with the motion controller to move the three-axis servo module in the X direction. During the movement, the three-axis servo module sends a position signal with a frequency of 1KHz to the trigger controller at intervals. The position signals are counted and numbered as [1, 2, 3…]. After receiving the signals, the trigger controller controls each component in the optical detection system respectively;

[0023] 2) The trigger controller sends 8 trigger pulses with a frequency of 20KHz to the LED stripe light source, and by setting the brightness of the LED lamp beads inside the LED stripe light source, the LED stripe light source is made to flash out 4 horizontal and 4 vertical sinusoidal phase-shifted stripe images at high speed. At the same time, the trigger controller sends 8 trigger pulses with a frequency of 20KHz to the line array camera, so that the line array camera collects 4 horizontal phase-shifted images and 4 vertical phase-shifted images of deflectometry within the time when the LED stripe light source is turned on, for a total of 8 lines of images;

[0024] 3) The trigger controller sequentially sends trigger pulses with a frequency of 20KHz to the diffuse reflection light source, so that the diffuse reflection light source is sequentially lit. At the same time, the trigger controller sequentially sends 4 trigger pulses with a frequency of 20KHz to the line array camera, so that the line array camera sequentially collects 4 lines of corresponding images of the photometric stereo method during the time when the diffuse reflection light source is sequentially lit;

[0025] 4) The line array camera completes the image acquisition of one position according to step 2) and step 3), with a total of 12 lines of images, and then waits for the next position signal of the three-axis servo module and the trigger pulse of the trigger controller;

[0026] 5) In the X-axis position signal with a frequency of 1 kHz and numbered 1, after the line array camera completes image acquisition according to step 2) and step 3), the trigger controller sends a trigger pulse with a frequency of 125 Hz to the two Y-direction line lasers in the Y-direction ranging and color difference detection component to light up the Y-direction line lasers. At the same time, the trigger controller sends a trigger pulse with a frequency of 250 Hz to the Y-direction area array camera in the Y-direction ranging and color difference detection component to enable the Y-direction area array camera to collect images corresponding to the Y-direction laser triangulation method within the time when the Y-direction line lasers are lit;

[0027] 6) In the X-axis position signal with a frequency of 1 KHz and numbered 2, after the line array camera completes image acquisition according to step 2) and step 3), the trigger controller sends a trigger pulse with a frequency of 125 Hz to the color difference detection light source in the Y-direction ranging and color difference detection component to light up the color difference detection light source. At the same time, the trigger controller sends a trigger pulse with a frequency of 250 Hz to the Y-direction area array camera in the Y-direction ranging and color difference detection component to enable the Y-direction area array camera to acquire RGB color images;

[0028] 7) In the X-axis position signal with a frequency of 1 KHz and numbered 3, after the line array camera completes image acquisition according to step 2) and step 3), the trigger controller sends a trigger pulse with a frequency of 1 Hz to the X-direction line laser to light up the X-direction line laser. At the same time, the trigger controller sends a trigger pulse with a frequency of 1 Hz to the X-direction area array camera to enable the X-direction area array camera to collect images corresponding to the X-direction laser triangulation method;

[0029] 8) In the X-axis position signals numbered 9, 17, 25 ... 8*124+1 with a frequency of 1 kHz, after the line array camera completes image acquisition according to step 2) and step 3), the trigger controller sends a trigger signal according to step 5) to enable the optical detection system to acquire the image corresponding to the Y-axis laser triangulation method;

[0030] 9) In the X-axis position signals numbered 10, 18, 26 ... 8*124+2 with a frequency of 1 KHz, after the line array camera completes image acquisition according to step 2) and step 3), the trigger controller sends a trigger signal according to step 6) to enable the optical detection system to acquire an RGB color image;

[0031] 10) For the X-axis position signal corresponding to the number not mentioned in the above steps, after the line array camera completes the image acquisition according to step 2) and step 3), the trigger controller does not send a trigger signal to each component in the optical detection system;

[0032] 11) After the X-axis position signals numbered 1 to 1000 are collected according to the above steps, a collection cycle is completed. In one cycle, one image corresponding to the X-axis laser triangulation method is collected, 125 images corresponding to the Y-axis laser triangulation method are collected, and 125 RGB color images are collected. The linear array camera collects 12,000 lines of images;

[0033] 12) Entering a subsequent image acquisition cycle, repeating steps 5) to 11) to acquire images, wherein in the X-axis position signals numbered 1003, 2003, 3003 ... 1000*n+3 with a frequency of 1 kHz, the optical detection system acquires the image corresponding to the X-axis laser triangulation method according to step 7);

[0034] 13) The X-axis laser triangulation corresponding image collected by the X-axis ranging component records the height information of the surface of the detected object along the X-axis direction. When the height change and normal change of the surface of the detected object exceed the set threshold, the signal is fed back to the industrial computer, and the industrial computer and the motion controller send a signal to the three-axis servo module, and the three-axis servo module adjusts the position of the optical detection system relative to the detected object on the Z axis and the RY axis of the coordinate system.

[0035] The steps for processing the acquired original image are as follows:

[0036] 1) At each image acquisition position, the original image acquired by the linear array camera has a total of height rows, i.e., [0, 1, 2, 3, ... (height-1)], which is split into 12 images by the following splitting method:

[0037] Image1=12*[0,1,2,3,4...(height / 12)-1]

[0038] Image2=12*[0, 1, 2, 3, 4...(height / 12)-1]+1

[0039] Image3=12*[0, 1, 2, 3, 4…(height / 12)-1]+2

[0040] …

[0041] Image12=12*[0, 1, 2, 3, 4…(height / 12)-1]+11

[0042] Among them, Image1 to Image4 are the horizontal phase shift images of the deflectometry, Image5 to Image8 are the vertical phase shift images of the deflectometry, and Image9 to Image12 are the corresponding images of the photometric stereo method, a total of 12 images;

[0043] 2) Calculate Image1 to Image12 to obtain the glossiness image at the incident angle, and extract the diffuse reflection illumination image, the specular reflection illumination image, and the albedo image in the image;

[0044] 3) According to the four-step phase shift principle, a lateral wrapped phase map is calculated through the four lateral phase shift images of the deflectometry, a vertical wrapped phase map is calculated through the four vertical phase shift images of the deflectometry, and a normal vector of the surface of the object under test is calculated through the four corresponding images of the photometric stereo method, and decomposed into a photometric stereo row gradient map and a photometric stereo column gradient map;

[0045] 4) calculating the triangulation height map of the Y-axis laser triangulation corresponding image through the laser triangulation principle and calibration information;

[0046] 5) stitching the collected multiple RGB color images to obtain a color image of the surface of the object to be tested, converting the color image into LAB space to form a color difference map, and detecting color-related defects;

[0047] 6) performing deflectometry unwrapping operation on the lateral wrapped phase map, the vertical wrapped phase map, the photometric stereo row gradient map, the photometric stereo column gradient map, and the triangulation height map to obtain a lateral phase map and a vertical phase map;

[0048] 7) Calculating the deflectometry row gradient map, deflectometry column gradient map and curvature map of the surface normal vector of the object under test in the phase measurement deflectometry system composed of the linear array camera and the LED stripe light source in combination with the calibration information;

[0049] 8) According to the above image analysis, 9 synthetic images are obtained, namely, gloss image, specular reflection illumination image, diffuse reflection illumination image, albedo image, color difference image, triangulation height image, deflectometry row gradient image, deflectometry column gradient image and curvature image. Through traditional image processing methods such as Blob analysis, edge extraction, template matching, color space conversion, frequency domain analysis and deep learning algorithms, abnormal area detection is performed on each synthetic image to obtain the suspected defect area of ​​the inspected object;

[0050] 9) For 2D defects of mirror-like objects, such as color difference defects, dirt defects, and texture defects, the suspected defect area of ​​the inspected object will be judged according to the 2D defect standard;

[0051] 10) For 3D defects of mirror-like objects, such as scratch and bump defects, and concave and convex defects, the triangulation height map, deflectometry row gradient map, and deflectometry column gradient map of the suspected defect area of ​​the inspected object are extracted separately, and three-dimensional reconstruction is performed in combination with the calibration information to obtain a relative height map to determine whether the relative height change of the area meets the 3D defect standard.

[0052] The beneficial effects of the present invention are as follows: the defect detection device for mirror-like objects, by designing an industrial computer, a trigger controller, a motion controller and a three-axis servo module for use with an optical detection system, can efficiently and quickly obtain directly or by calculation the mirror reflection map, diffuse reflection map, gloss map, albedo map, color difference map, height map, gradient map, curvature map, etc. of the surface of the mirror-like object; thereby, the color difference defects, dirt defects, texture defects, scratch and bump defects, concave and convex defects, etc. of the mirror-like object can be detected by traditional image processing algorithms and deep learning algorithms; the present invention can detect many types of defects, and can realize color difference defects, dirt defects, Simultaneous detection of texture defects, scratch and bump defects, and concave and convex defects; wide range of applications, from pure mirror surfaces with high gloss to mirror-like surfaces with low gloss, and can accommodate changes in the material of the object surface; high detection accuracy, capable of achieving micron-level defect detection accuracy; strong adaptability, able to adapt to changes in object height and normal during scanning; high detection efficiency, the deflectometry system only needs to collect 8 phase-shifted images; at the same time, it also has the advantages of compact structure, low system cost, and large field of view of linear array cameras, and is suitable for defect detection of large and complex curved products and cylindrical products, such as car body paint, industrial glass, ceramics, large metal processing parts and other products. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a schematic diagram of the overall structure of the present invention;

[0054] Figure 2 It is a schematic diagram of the diffuse reflection light source structure of the present invention;

[0055] Figure 3 It is a timing diagram of image acquisition control of the present invention;

[0056] Figure 4 It is a schematic diagram of the principle of phase shift and phase unwrapping in the present invention;

[0057] Figure 5 The coordinate system of the deflectometry system in the present invention is defined;

[0058] Figure 6This is a schematic diagram of imaging principle on the XOZ plane of the deflectometry of the linear array camera in the present invention;

[0059] Figure 7 This is a schematic diagram of imaging principle on the YOZ plane of the deflectometry of the linear array camera in the present invention;

[0060] Figure 8 Schematic diagram of the relationship between the original image and the synthesized image in the present invention. DETAILED DESCRIPTION

[0061] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making a clearer and more definite definition of the protection scope of the present invention.

[0062] like Figure 1 The defect detection device for a mirror-like object shown in the figure comprises an optical detection system 1 , an industrial computer 2 , a trigger controller 3 , a motion controller 4 and a three-axis servo module 5 .

[0063] The industrial computer 2 is connected to the motion controller 4. The three-axis servo module 5 can drive the optical detection system 1 to scan the object to be detected 9 under the action of the industrial computer 2 and the motion controller 4. The trigger controller 3 can be connected to the three-axis servo module 5 and the optical detection system 1 respectively, and the optical detection system 1 is connected to the industrial computer 2.

[0064] The optical detection system 1 is used to collect and analyze the surface data of mirror-like objects, and includes a line array camera 11, an LED stripe light source 12, four diffuse reflection light sources 13, a Y-direction distance measurement and color difference detection component 14 and an X-direction distance measurement component 15.

[0065] The line array camera 11 and the LED stripe light source 12 are symmetrical about the surface normal of the detected object 9 , and the angle between the line array camera 11 and the LED stripe light source 12 is 10° to 60°.

[0066] The four diffuse reflection light sources 13 are evenly distributed around the surface normal of the detected object 9, and the angle between the diffuse reflection light source 13 and the surface normal of the detected object 9 is 60° to 80°. Figure 2 As shown, the diffuse reflection light source 13 includes a cooling fan 131, a heat conductive material 132, a patch LED 133, a focusing rod 134 and a diffusion film 135. The light of the patch LED 133 is inclined to emit from the light source surface at an angle of 40° to 50°.

[0067] The Y-direction ranging and color difference detection component 14 includes two Y-direction line lasers 141, a color difference detection light source 142 and a Y-direction area array camera 143. The two Y-direction line lasers 141 are coplanar with the line array camera 11 and are distributed on both sides of the line array camera 11. The Y-direction area array camera 143 and the Y-direction line lasers 141 are respectively arranged on both sides of the surface normal of the detected object 9. The angle between the Y-direction area array camera 143 and the surface normal of the detected object 9 is 40° to 60°. The Y-direction area array camera 143 and the color difference detection light source 142 are symmetrical about the surface normal of the detected object 9.

[0068] The X-direction ranging component 15 includes an X-direction line laser 151 and an X-direction area array camera 152. The X-direction line laser 151 is perpendicular to the YOZ plane of the coordinate system. The X-direction area array camera 152 and the X-direction line laser 151 are respectively arranged on both sides of the surface normal of the detected object 9. The angle between the X-direction area array camera 152 and the surface normal of the detected object 9 is 40° to 60°.

[0069] When using this type of mirror object defect detection device for image acquisition, such as Figure 3 As shown, the following steps are included:

[0070] 1) The industrial computer 2 communicates with the motion controller 4 to move the three-axis servo module 5 in the X direction. During the movement, the three-axis servo module 5 sends a position signal with a frequency of 1KHz to the trigger controller 3 at intervals. The position signals are counted and numbered as [1, 2, 3...]. After receiving the signals, the trigger controller 3 controls the various components in the optical detection system 1 respectively;

[0071] 2) The trigger controller 3 sends 8 trigger pulses with a frequency of 20KHz to the LED stripe light source 12, and by setting the brightness of the LED lamp beads inside the LED stripe light source 12, the LED stripe light source 12 is made to flash 4 horizontal and 4 vertical sinusoidal phase-shifted stripe images at high speed. At the same time, the trigger controller 3 sends 8 trigger pulses with a frequency of 20KHz to the line array camera 11, so that the line array camera 11 collects 4 horizontal phase-shifted images and 4 vertical phase-shifted images of deflectometry within the time when the LED stripe light source 12 is turned on, for a total of 8 lines of images;

[0072] 3) The trigger controller 3 sequentially sends trigger pulses with a frequency of 20 KHz to the diffuse reflection light source 13, so that the diffuse reflection light source 13 is sequentially lit. At the same time, the trigger controller 3 sequentially sends 4 trigger pulses with a frequency of 20 KHz to the line array camera 11, so that the line array camera 11 sequentially collects 4 lines of corresponding images of the photometric stereo method during the time when the diffuse reflection light source 13 is sequentially lit.

[0073] 4) The line array camera 11 completes the image acquisition of one position according to step 2) and step 3), with a total of 12 lines of images, and then waits for the next position signal of the three-axis servo module 5 and the trigger pulse of the trigger controller 3;

[0074] 5) In the X-axis position signal with a frequency of 1 kHz and numbered 1, after the line array camera 11 completes image acquisition according to step 2) and step 3), the trigger controller 3 sends a trigger pulse with a frequency of 125 Hz to the two Y-direction line lasers 141 in the Y-direction ranging and color difference detection component 14 to light up the Y-direction line lasers 141. At the same time, the trigger controller 3 sends a trigger pulse with a frequency of 250 Hz to the Y-direction area array camera 143 in the Y-direction ranging and color difference detection component 14, so that the Y-direction area array camera 143 collects the image corresponding to the Y-direction laser triangulation method within the time when the Y-direction line lasers 141 are lit;

[0075] 6) In the X-axis position signal with a frequency of 1 kHz and numbered 2, after the line array camera 11 completes image acquisition according to step 2) and step 3), the trigger controller 3 sends a trigger pulse with a frequency of 125 Hz to the color difference detection light source 142 in the Y-direction distance measurement and color difference detection component 14 to light up the color difference detection light source 142. At the same time, the trigger controller 3 sends a trigger pulse with a frequency of 250 Hz to the Y-direction area array camera 143 in the Y-direction distance measurement and color difference detection component 14 to enable the Y-direction area array camera 143 to acquire RGB color images;

[0076] 7) In the X-axis position signal with a frequency of 1 kHz and numbered 3, after the line array camera 11 completes image acquisition according to step 2) and step 3), the trigger controller 3 sends a trigger pulse with a frequency of 1 Hz to the X-direction line laser 151 to light up the X-direction line laser 151. At the same time, the trigger controller 3 sends a trigger pulse with a frequency of 1 Hz to the X-direction area array camera 152 to enable the X-direction area array camera 152 to collect images corresponding to the X-direction laser triangulation method;

[0077] 8) In the X-axis position signals numbered 9, 17, 25 ... 8*124+1 with a frequency of 1KHz, after the line array camera 11 completes image acquisition according to step 2) and step 3), the trigger controller 3 sends a trigger signal according to step 5) to enable the optical detection system 1 to acquire the image corresponding to the Y-axis laser triangulation method;

[0078] 9) In the X-axis position signals numbered 10, 18, 26 ... 8*124+2 with a frequency of 1KHz, after the line array camera 11 completes image acquisition according to step 2) and step 3), the trigger controller 3 sends a trigger signal according to step 6) to enable the optical detection system 1 to acquire an RGB color image;

[0079] 10) For the X-axis position signal corresponding to the number not mentioned in the above steps, after the line array camera 11 completes the image acquisition according to step 2) and step 3), the trigger controller 3 does not send a trigger signal to each component in the optical detection system 1;

[0080] 11) After the X-axis position signals numbered 1 to 1000 are collected according to the above steps, a collection cycle is completed. In one cycle, one image corresponding to the X-axis laser triangulation method is collected, 125 images corresponding to the Y-axis laser triangulation method are collected, and 125 RGB color images are collected. The linear array camera collects 12,000 lines of images;

[0081] 12) Entering a subsequent image acquisition cycle, repeating steps 5) to 11) to acquire images, wherein in the X-axis position signals with a frequency of 1 kHz and numbered 1003, 2003, 3003, ..., 1000*n+3, the optical detection system acquires the image corresponding to the X-axis laser triangulation method according to step 7);

[0082] 13) The X-axis laser triangulation corresponding image collected by the X-axis ranging component 15 records the height information of the surface of the detected object along the X-axis direction. When the height change and normal change of the surface of the detected object exceed the set threshold value, the signal is fed back to the industrial computer 2, and the industrial computer 2 and the motion controller 4 send a signal to the three-axis servo module 5, and the three-axis servo module 5 adjusts the position of the optical detection system 1 relative to the detected object on the Z axis and the RY axis of the coordinate system.

[0083] The origin of the world coordinate system is defined to be on the optical axis of the line array camera 11, as Figure 5 As shown, the origin of the coordinate system of the LED stripe light source 12 is on the optical axis reflection line. The coordinate systems of the linear array camera 11 and the LED stripe light source 12 are calibrated in advance, and the coordinate system of the LED stripe light source 12 can be obtained. s In the coordinate system O of the line array camera 11 c The rotation matrix R in cs and the translation matrix T cs , Linear camera 11 coordinate system O c In the world coordinate system O w The rotation matrix R cw and the translation matrix T cw , and the LED stripe light source 12 coordinate system O s In the world coordinate system O w The rotation matrix R sw and the translation matrix T sw For any point P(X w ,Y w ,Z w ) Linear camera 11 coordinate system Oc The coordinates of the c ,Z c ), in the LED stripe light source 12 coordinate system O s The coordinates below are (X s ,Y s ,Z s ).

[0084] Project the optical system onto the XOZ plane of the world coordinate system, such as Figure 6 As shown, from the geometric relationship, when the reference plane deflects θ x When the corresponding reflected light is deflected by 2θ x , point P before and after the deflection corresponds to point S and point S' on the light source respectively, and the periods of the horizontal and vertical sinusoidal fringes in the image are p cx and p cy , their phase information difference at the image point P(u,v) The calculation formula is as follows:

[0085]

[0086] Project the optical system onto the YOZ plane of the world coordinate system, such as Figure 7 As shown, from the geometric relationship, when the reference plane deflects θ y When the corresponding reflected light is deflected by 2θ y , before and after the deflection, the P point corresponds to the S and S' points on the light source respectively. The phase information difference between them at the image point P(u,v) is The calculation formula is as follows:

[0087]

[0088] From the above formula, we can see that p cx 、p cy , T sw , α, β, Y c All of them can be calculated from the calibration information. The one related to the defect information is tan(θ x ) and tan(θ y ), need to calculate and Z c To reduce the number of images collected by deflectometry, the present invention introduces photometric stereo to take 4 images to assist in solving Laser triangulation method is introduced to assist in solving Z c In the photometric stereo method, the illumination directions of the four diffuse reflection light sources 13 are calibrated in advance. During the calibration process, the normal vector of the line array camera 11 is calculated by R cw Obtained by conversion.

[0089] The steps for processing the acquired original image are as follows:

[0090] 1) At each image acquisition position, the original image acquired by the line array camera 11 has a total of height rows, namely [0, 1, 2, 3, ... (height-1)], which is split into 12 images by the following splitting method:

[0091] Image1=12*[0,1,2,3,4...(height / 12)-1]

[0092] Image2=12*[0, 1, 2, 3, 4...(height / 12)-1]+1

[0093] Image3=12*[0, 1, 2, 3, 4…(height / 12)-1]+2

[0094] …

[0095] Image12=12*[0, 1, 2, 3, 4…(height / 12)-1]+11

[0096] Among them, Image1 to Image4 are the horizontal phase shift images of the deflectometry, Image5 to Image8 are the vertical phase shift images of the deflectometry, and Image9 to Image12 are the corresponding images of the photometric stereo method, a total of 12 images;

[0097] 2) Calculate Image1 to Image12 to obtain the glossiness image at the incident angle, and extract the diffuse reflection illumination image, the specular reflection illumination image, and the albedo image in the image;

[0098] The calculation process is: 4 deflectometry horizontal phase shift images and 4 deflectometry vertical phase shift images, such as Figure 4 As shown, their sinusoidal phases differ by π / 2. The light intensity in the image collected by the linear array camera 11 is:

[0099] I1(u,v)=I dc +I a cos[φ(u,v)]

[0100] I2(u,v)=I dc +I a cos[φ(u,v)+π / 2]

[0101] I3(u,v)=I dc +I a cos[φ(u,v)+π]

[0102] I4(u,v)=I dc +I a cos[φ(u,v)+3π / 2]

[0103] 3) According to the four-step phase shift principle, a lateral wrapped phase map is calculated through the four lateral phase shift images of the deflectometry, a vertical wrapped phase map is calculated through the four vertical phase shift images of the deflectometry, and a normal vector of the surface of the object under test is calculated through the four corresponding images of the photometric stereo method, and decomposed into a photometric stereo row gradient map and a photometric stereo column gradient map;

[0104] In the four-step phase shift principle, the wrapped phase of the image The calculation formula is:

[0105]

[0106] Wrapping phase calculated by phase shifting algorithm is unique within one phase cycle, but due to the presence of multiple fringes throughout the measurement space, The distribution is sawtooth-shaped, and the phase main value of the spatial point must be phase-unwrapped to obtain a continuous absolute phase value. like Figure 4 As shown;

[0107] 4) By using the laser triangulation principle and calibration information, the Y-axis laser triangulation corresponding image is calculated to obtain a triangulation height map, which corresponds to the Z-axis in the deflectometry system. c ;

[0108] 5) stitching the collected multiple RGB color images to obtain a color image of the surface of the object to be tested, converting the color image into LAB space to form a color difference map, and detecting color-related defects;

[0109] 6) performing deflectometry unwrapping operation on the lateral wrapped phase map, the vertical wrapped phase map, the photometric stereo row gradient map, the photometric stereo column gradient map, and the triangulation height map to obtain a lateral phase map and a vertical phase map;

[0110] 7) Calculating the deflectometry row gradient map, deflectometry column gradient map and curvature map of the surface normal vector of the object under test in the phase measurement deflectometry system composed of the linear array camera 11 and the LED stripe light source 12 in combination with the calibration information;

[0111] 8) According to the above image analysis, 9 synthetic images are obtained, namely, gloss image, specular reflection illumination image, diffuse reflection illumination image, albedo image, color difference image, triangulation height image, deflectometry row gradient image, deflectometry column gradient image and curvature image. Through traditional image processing methods such as Blob analysis, edge extraction, template matching, color space conversion, frequency domain analysis and deep learning algorithms, abnormal area detection is performed on each synthetic image to obtain the suspected defect area of ​​the inspected object;

[0112] After the suspected defect area is detected through the above image, it is often necessary to quantitatively analyze the height of the suspected area to determine whether the relative height change of the area meets the defect standard. In order to reduce the amount of calculation, the image data of the suspected area can be extracted separately, and the height of the area can be accurately reconstructed through the gradient and triangulation height map to obtain a relative height map;

[0113] 9) For 2D defects of mirror-like objects, such as color difference defects, dirt defects, and texture defects, the suspected defect area of ​​the inspected object will be judged according to the 2D defect standard;

[0114] 10) For 3D defects of mirror-like objects, such as scratch and bump defects, and concave and convex defects, the triangulation height map, deflectometry row gradient map, and deflectometry column gradient map of the suspected defect area of ​​the inspected object are extracted separately, and three-dimensional reconstruction is performed in combination with the calibration information to obtain a relative height map to determine whether the relative height change of the area meets the 3D defect standard.

[0115] In the description of the present invention, it should be noted that the orientations or positional relationships indicated by terms such as “upper”, “lower”, “left”, “right”, “top”, “bottom”, “side” and “end” are based on the orientations or positional relationships shown in the accompanying drawings. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, they cannot be understood as limitations on the present invention.

[0116] In addition, in the description of the embodiments of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "set", "provided with", etc. should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0117] Finally, it should be noted that the above embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A defect detection device for a mirror-like object, characterized in that: The optical detection system comprises an optical detection system, an industrial computer, a trigger controller, a motion controller and a three-axis servo module. The industrial computer is connected to the motion controller. The three-axis servo module can drive the optical detection system to scan the detected object under the action of the industrial computer and the motion controller. The trigger controller can be connected to the three-axis servo module and the optical detection system respectively. The optical detection system is connected to the industrial computer. The optical detection system is used to collect and analyze the surface data of mirror-like objects, including a line array camera, an LED stripe light source, a plurality of diffuse reflection light sources, a Y-direction distance measurement and color difference detection component and an X-direction distance measurement component. The line array camera and the LED stripe light source are symmetrical about the surface normal of the detected object, and the diffuse reflection light source is evenly distributed around the surface normal of the detected object. The Y-direction distance measurement and color difference detection component includes two Y-direction line lasers, a color difference detection light source and a Y-direction area array camera. The two Y-direction line lasers are coplanar with the line array camera and are distributed on both sides of the line array camera. The Y-direction area array camera and the Y-direction line laser are respectively arranged on both sides of the surface normal of the detected object. The Y-direction area array camera and the color difference detection light source are symmetrical about the surface normal of the detected object. The X-direction distance measurement component includes an X-direction line laser and an X-direction area array camera. The X-direction line laser is perpendicular to the YOZ plane of the coordinate system. The X-direction area array camera and the X-direction line laser are respectively arranged on both sides of the surface normal of the detected object. The angle between the linear array camera and the LED stripe light source is 10°~60°, the angle between the diffuse reflection light source and the surface normal of the detected object is 60°~80°, the angle between the Y-axis area array camera and the surface normal of the detected object is 40°~60°, and the angle between the X-axis area array camera and the surface normal of the detected object is 40°~60°. The diffuse reflection light source comprises a cooling fan, a heat conducting material, a patch LED, a focusing rod and a diffusion film. The light of the patch LED is emitted obliquely from the light source surface at an angle of 40° to 50°.

2. The defect detection device for mirror-like objects according to claim 1, characterized in that: There are four diffuse reflection light sources.

Images