Light guide plate defect analysis method and system

Abstract

This invention discloses a method and system for analyzing defects in light guide plates, relating to the field of light guide plate defect analysis technology. The method includes the following steps: setting up a defect scanning workstation, irradiating the light guide plate under different wavelength conditions, and capturing initial defect images; preprocessing the defect recognition images to obtain defect recognition images; performing defect analysis on the defect recognition images, extracting defect boundaries from the defect recognition images, and obtaining defect regions from the defect boundaries; naming the light guide plate with defect regions a test plate, and performing a perception test on the test plate, testing and analyzing to obtain a perception threshold; extracting defect parameters from the defect regions, and judging whether the defects in the light guide plate have no impact based on the perception threshold. This invention addresses the problem that existing light guide plate defect analysis technologies still rely excessively on manually set thresholds and have inaccurate defect judgments, leading to low yield rates for light guide plates.

Description

Technical Field

[0001] This invention relates to the field of light guide plate defect analysis technology, specifically to a method and system for analyzing light guide plate defects. Background Technology

[0002] Light guide plate defect analysis technology refers to a comprehensive quality inspection and process monitoring technology that integrates optical imaging, image processing and intelligent algorithms to automatically detect, identify, classify, evaluate and trace the root causes of various defects generated during the production process of light guide plates.

[0003] Existing light guide plate defect analysis technologies often rely excessively on manual inspection, and during the inspection process, they depend heavily on manually set thresholds without scientific analysis of these thresholds. Furthermore, these technologies struggle to identify minute defects in light guide plates, some of which are invisible to the naked eye and do not affect normal use. However, existing technologies typically define all defective light guide plates as unqualified, leading to a significant drop in yield. For example, patent application CN114994071A discloses a "light guide plate inspection method and light guide plate inspection fixture," which relies excessively on manually set values ​​during inspection, lacking scientific analysis of these values ​​and failing to ensure their rationality and effectiveness. These existing light guide plate defect analysis technologies also suffer from excessive reliance on manually set thresholds and inaccurate defect assessments, resulting in low yield rates for light guide plates. Summary of the Invention

[0004] This invention aims to at least partially solve one of the technical problems in the prior art. By constructing a defect scanning workbench, configuring an adjustable wavelength light source and an industrial camera, and then acquiring initial defect images and spectral maps through the defect scanning workbench, the defect identification images are preprocessed to obtain defect identification images. Defect analysis is then performed on the defect identification images to extract defect boundaries and obtain defect regions from the defect boundaries. The light guide plate with defect regions is then named a test plate, and a perception test is performed on the test plate. The perception threshold is obtained through testing and analysis. Finally, defect parameters of the defect regions are extracted, and the perception threshold is used to determine whether the defects of the light guide plate have no impact. This invention solves the problem that existing light guide plate defect analysis technologies still rely too much on manually set thresholds and have inaccurate defect judgments, resulting in low yield rates for light guide plates.

[0005] To achieve the above objectives, in a first aspect, this application provides a method for analyzing defects in a light guide plate, comprising the following steps:

[0006] A defect scanning workbench was set up, and the light guide plate was irradiated under different wavelength conditions to obtain initial images of the defects.

[0007] Image preprocessing is performed on the defect identification image to obtain the defect identification image;

[0008] Defect analysis is performed on the defect recognition image to extract the defect boundary in the defect recognition image, and the defect region is obtained from the defect boundary;

[0009] The light guide plate with defective areas is named the test plate, and a perception test is performed on the test plate to obtain the perception threshold through testing and analysis.

[0010] Extract the defect parameters of the defect area and refer to the perception threshold to determine whether the defects of the light guide plate have no impact.

[0011] Furthermore, a defect scanning workbench is set up, and the light guide plate is irradiated under different wavelength conditions to obtain initial images of the defects, including the following sub-steps:

[0012] Set up a defect scanning workbench and equip it with an adjustable wavelength light source and an industrial camera;

[0013] Initial images and spectral diagrams of defects are acquired using a defect scanning workbench.

[0014] Furthermore, the process of setting up a defect scanning workbench and configuring an adjustable wavelength light source and an industrial camera includes the following sub-steps:

[0015] Construct a closed, dark space and name it the collection room. Name the side of the collection room the light source side, which includes the left side, front side, right side, and rear side.

[0016] An industrial camera is installed at the geometric center of the top of the acquisition chamber;

[0017] Connect the midpoints of the upper and lower edges of the light source surface to obtain the mounting line. Install the slide rail along the mounting line and mount the adjustable wavelength light source on the slide rail.

[0018] The adjustable wavelength light source is always aligned with the geometric center of the bottom of the acquisition chamber. The adjustable wavelength light source is moved in ascending order, and a flawed light guide plate is placed at the geometric center of the bottom of the acquisition chamber to capture initial images of defects in real time. The initial image of the defect with the most obvious flaw is manually selected as the reference image. The position of the adjustable wavelength light source corresponding to the reference image is marked as the installation point, and the adjustable wavelength light source is fixed at the installation point. The acquisition chamber, the adjustable wavelength light source, and the industrial camera together form a defect scanning workbench.

[0019] Furthermore, acquiring initial defect images and spectral maps via a defect scanning workbench includes the following sub-steps:

[0020] The light guide plate is fed into the defect scanning stage, and the geometric center of the light guide plate overlaps with the geometric center of the bottom of the defect scanning stage.

[0021] The adjustable wavelength light sources installed on the left side, front side, right side and rear side are named left light source, front light source, right light source and rear light source respectively. The light guide plate is illuminated in the order of left light source, front light source, right light source and rear light source, and only one adjustable wavelength light source lights up at a time.

[0022] Each time the adjustable wavelength light source is turned on, the wavelength of the adjustable wavelength light source is adjusted sequentially in order of increasing wavelength, and the industrial camera continuously records the image of each frame to obtain the recognition frame image;

[0023] Acquire the recognition frame image, perform grayscale processing on the recognition frame image, obtain the grayscale value of each pixel, and name it as single-point grayscale. When analyzing any pixel, name it as the point to be analyzed, name the grayscale value of the point to be analyzed as the grayscale value to be analyzed, name the pixels in the eight neighborhoods of the point to be analyzed as the neighboring points to be analyzed, name the grayscale values ​​of the neighboring points to be analyzed as the neighboring point grayscale values, calculate the absolute value of the difference between the grayscale value to be analyzed and the grayscale value of each neighboring point, name it as the difference value to be analyzed, obtain the maximum value among the differences to be analyzed, name it as the maximum difference value, calculate the maximum difference value of each pixel's low point, and then calculate the sum of all the maximum differences, name it as the grayscale difference degree.

[0024] Obtain the recognition frame image with the largest grayscale difference and name it the initial defect image.

[0025] Furthermore, the image preprocessing for the defect identification image specifically involves image enhancement and grayscale processing to ultimately obtain the defect identification image.

[0026] Further, defect analysis is performed on the defect recognition image to extract the defect boundaries and obtain the defect region from the defect boundaries, including the following sub-steps:

[0027] The defect identification images captured under illumination from the left light source, front light source, right light source, and rear light source are named the left defect identification image, front defect identification image, right defect identification image, and rear defect identification image, respectively. When analyzing any defect identification image, it is named the target analysis image.

[0028] The pixels in the target analysis image are named target image points, the gray values ​​of the target image points are named target image gray values, the target image point in the i-th row and j-th column of the target analysis image is labeled as P(i,j), where i and j are both non-zero natural numbers and (i,j) is the index of P, and the target image gray value of P(i,j) is labeled as H(i,j).

[0029] Label each row's P(i,j) as PH in top-to-bottom order, working by row. i In each column, P(i,j) is labeled as PL in order from left to right. j ;

[0030] If the target analysis map is a left defect identification map or a right defect identification map, then obtain PH. i At the same time, pH i The P(i,j) in the image are numbered and represented by the symbol TH(i,n), where n is a non-zero natural number and the range of n is the same as that of j. The smaller n is in TH(i,n), the closer the target image point is to the lit tunable wavelength light source.

[0031] If the target analysis diagram is a pre-defect identification diagram or a post-defect identification diagram, then obtain PL. j At the same time, for PL j The P(i,j) in the image are numbered and represented by the symbol TL(j,m), where m is a non-zero natural number and the range of m is the same as that of i. The smaller m is in TL(j,m), the closer the target image point is to the lit tunable wavelength light source.

[0032] Label TH(i,n) and TL(j,m) as TR(h,g), where h and g are both non-zero natural numbers and (h,g) is the index of TR. When the target analysis image is a left defect recognition image or a right defect recognition image, TR(h,g) represents TH(i,n). When the target analysis image is a front defect recognition image or a back defect recognition image, TR(h,g) represents TL(j,m). Label the grayscale of the target image corresponding to TR(h,g) as TE(h,g).

[0033] Analyze each value of h independently. Starting with g=2, calculate |TE(h,g)-TE(h,g-1)|, mark the calculation result as CL(h,g), increment g and repeat the calculation of CL(h,g) until the maximum value of g is reached.

[0034] A one-dimensional coordinate system is established with CL(h,g) as the X-axis and named the preliminary defect analysis diagram. CL(h,g) is entered into the preliminary defect analysis diagram according to g. Cluster analysis is performed on the preliminary defect analysis diagram using a clustering algorithm to obtain difference clusters. The number of CL(h,g) in the difference clusters is counted and named the cluster size. The difference cluster with the largest cluster size is removed. The remaining difference clusters are collectively referred to as the defect to be confirmed clusters. TR(h,g) corresponding to CL(h,g) in the defect to be confirmed clusters are named the defect to be confirmed points.

[0035] Further analysis is performed on the defect points to be confirmed to obtain the defect boundary, and the defect area is obtained from the defect boundary.

[0036] Further analysis is performed on the defect points to be confirmed to obtain the defect boundaries, and the defect region is derived from the defect boundaries, including the following sub-steps:

[0037] Analyze the defect points to be confirmed for each value of h. If any two defect points to be confirmed are adjacent in the target analysis diagram, they are integrated into a defect line to be confirmed. If any defect point to be confirmed is adjacent to any defect line to be confirmed, the defect point to be confirmed is included in the defect line to be confirmed.

[0038] After analyzing the target analysis map, different defect confirmation lines are obtained. The defect confirmation lines obtained from the analysis of the left defect identification map, the front defect identification map, the right defect identification map, and the rear defect identification map are colored in the target analysis map. All target image points on the defect confirmation lines are filled with pure white to obtain the defect boundaries.

[0039] Extract the closed region enclosed by the defect boundary and name it the defect region.

[0040] Furthermore, the light guide plate with defective areas is named the test plate, and a perception test is performed on the test plate. The perception threshold is obtained by testing and analysis, including the following sub-steps:

[0041] The light guide plate with defective areas is named the test plate. The test plate is used for light guiding. The tester observes the test plate and judges whether there is a defect. If the test plate is judged to have a defect, it is marked as a defective plate. If the test plate is judged not to have a defect, it is marked as a normal plate.

[0042] The defective area in the defective board is named the sensitive area, and the pixel in the sensitive area is named the sensitive point. The defective area in the normal board is named the non-sensitive area, and the pixel in the non-sensitive area is named the non-sensitive point.

[0043] Get the difference of j between P(i,j) corresponding to the same row of the sensed points in the sensed region, named the sensed row width; get the difference of i between P(i,j) corresponding to the same column of the sensed points in the sensed region, named the sensed column width; get the maximum value of the sensed row width and sensed column width, named the single area width; get the minimum value of the single area width in the sensed regions of all defective boards, named the minimum sensed defect width.

[0044] Get the difference of j between P(i,j) corresponding to the same row of the non-sensitive points in the non-sensitive area, and name it as the non-sensitive row width. Get the difference of i between P(i,j) corresponding to the same column of the non-sensitive points in the non-sensitive area, and name it as the non-sensitive column width. Get the maximum value of the non-sensitive row width and non-sensitive column width of the non-sensitive area of ​​all defective boards, and name it as the maximum non-sensitive defect width.

[0045] The average of the minimum perceptible defect width and the maximum non-perceptible defect width is used to obtain the perception threshold.

[0046] Further, extracting defect parameters from the defective area and determining whether the defects in the light guide plate have no impact by referring to the perception threshold includes the following sub-steps:

[0047] Name the light guide plate that needs to be analyzed as the plate to be analyzed, and extract the defect area of ​​the plate to be analyzed and name it as the area to be analyzed.

[0048] Extract the width of a single region in the area to be analyzed and name it the defect parameter. Compare the defect parameter with the defect threshold. If the defect parameter is less than the defect threshold, output a surface defect signal; otherwise, output a surface qualified signal.

[0049] If the output surface fails to meet the requirements, it indicates that the light guide plate is defective, meaning the light guide plate is unqualified.

[0050] Secondly, this application provides a light guide plate defect analysis system, including a defect acquisition module, an image preprocessing module, a region recognition module, a perception analysis module, and an impact judgment module; the defect acquisition module, the image preprocessing module, the perception analysis module, and the impact judgment module are respectively data connected to the region recognition module.

[0051] The defect acquisition module is used to build a defect scanning workbench, irradiate the light guide plate under different wavelength conditions, and capture initial images of the defects.

[0052] The image preprocessing module is used to preprocess the defect identification image to obtain the defect identification image;

[0053] The region recognition module is used to perform defect analysis on the defect recognition image, extract the defect boundary in the defect recognition image, and obtain the defect region from the defect boundary;

[0054] The perception analysis module is used to name the light guide plate with defective areas as the test plate, and to perform perception tests on the test plate to obtain the perception threshold through testing and analysis.

[0055] The impact judgment module is used to extract the defect parameters of the defect area and determine whether the defects of the light guide plate have no impact by referring to the perception threshold.

[0056] The beneficial effects of this invention are as follows: This invention constructs a defect scanning workbench, configures an adjustable wavelength light source and an industrial camera, and then acquires initial defect images and spectral maps through the defect scanning workbench. The defect recognition images are preprocessed to obtain defect recognition images, and then defect analysis is performed on the defect recognition images to extract the defect boundaries and obtain the defect regions from the defect boundaries. The advantage is that there are no artificially set thresholds in the detection process, and all values ​​are obtained through testing. In addition, during the analysis process, light of different wavelengths at a certain angle is used to illuminate the light guide plate, so that the defect parts on the light guide plate appear with certain shadows. These shadows are then analyzed to form the defect regions. If traditional white light is used for illumination, the defect parts may not be obvious due to the long wavelength, which further increases the difficulty of image recognition. This invention improves the accuracy and rationality of light guide plate defect analysis.

[0057] This invention names a light guide plate with defective areas as a test plate, performs a perception test on the test plate, analyzes the test to obtain the perception threshold, and finally extracts the defect parameters of the defective areas. The perception threshold is then used to determine whether the defects of the light guide plate have no impact. The advantage is that it can further analyze the perception threshold to determine whether the defects will affect the normal use of the light-blocking plate, thereby improving the yield of the light-blocking plate and improving the accuracy and effectiveness of the light guide plate defect analysis. Attached Figure Description

[0058] Figure 1 This is a schematic diagram of the system of the present invention;

[0059] Figure 2 This is a schematic diagram of the target analysis diagram of the present invention;

[0060] Figure 3 This is a schematic diagram of the grayscale of a portion of the target image according to the present invention;

[0061] Figure 4 This is a magnified schematic diagram of the defect boundary of the present invention;

[0062] Figure 5 This is a flowchart of the steps of the method of the present invention. Detailed Implementation

[0063] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] Example 1, please refer to Figure 1As shown, this application provides a light guide plate defect analysis system, including a defect acquisition module, an image preprocessing module, a region recognition module, a perception analysis module, and an impact judgment module; the defect acquisition module, image preprocessing module, perception analysis module, and impact judgment module are respectively connected to the region recognition module for data analysis.

[0065] The defect acquisition module is used to build a defect scanning workbench, irradiate the light guide plate under different wavelength conditions, and capture initial images of the defects; the defect acquisition module includes a workbench building unit and a defect scanning unit.

[0066] The workbench assembly unit is used to build a defect scanning workbench and is equipped with an adjustable wavelength light source and an industrial camera.

[0067] The workbench building unit is configured with workbench building strategies, which include:

[0068] Construct a closed, dark space and name it the Collection Room. Name the sides of the Collection Room the Light Source Surfaces, which include the left side, front side, right side, and rear side.

[0069] An industrial camera is installed at the geometric center of the top of the acquisition chamber;

[0070] Connect the midpoints of the upper and lower edges of the light source surface to obtain the mounting line. Install the slide rail along the mounting line and mount the adjustable wavelength light source on the slide rail.

[0071] The tunable wavelength light source is always aligned with the geometric center of the bottom of the acquisition chamber. The tunable wavelength light source is moved in order from low to high. A flawed light guide plate is placed at the geometric center of the bottom of the acquisition chamber to capture initial images of the defects in real time. The initial image of the defect with the most obvious flaw is manually selected as the reference image. The position of the tunable wavelength light source corresponding to the reference image is marked as the installation point. The tunable wavelength light source is fixed at the installation point. The acquisition chamber, the tunable wavelength light source, and the industrial camera together form the defect scanning workbench.

[0072] In practical applications, a closed, dark space is used as the acquisition room to prevent other light sources from interfering with the shooting results. The sliding rail is installed to adjust the angle at which the adjustable wavelength light source illuminates the light guide plate. The adjustable wavelength light source is ultimately aligned with the geometric center of the bottom of the acquisition room. In fact, the adjustable wavelength light source must be directly facing the geometric center of the bottom of the acquisition room and must not be offset to the surrounding areas. Since the scratches on the light guide plate are actually small grooves on the surface of the light guide plate, when the adjustable wavelength light source is illuminating at a certain incident angle, there will be some unilluminated areas in the grooves, which will produce a certain shadow. Then, an industrial camera is used for shooting. The industrial camera uses ultra-high resolution and can identify subtle color differences. Since the wavelength of light is different, the degree of shadow is also different. Therefore, the wavelength needs to be dynamically adjusted to find the most obvious shadow. When determining the installation point, since only one-time debugging is required, manual judgment can be used directly. In actual use, machine recognition can also be used, which is similar to the grayscale difference in the defect scanning strategy.

[0073] The defect scanning unit is used to acquire initial images and spectral maps of defects via the defect scanning workbench;

[0074] The defect scanning unit is configured with a defect scanning strategy, which includes:

[0075] The light guide plate is fed into the defect scanning stage, and the geometric center of the light guide plate overlaps with the geometric center of the bottom of the defect scanning stage.

[0076] The adjustable wavelength light sources installed on the left side, front side, right side and rear side are named left light source, front light source, right light source and rear light source respectively. The light guide plate is illuminated in the order of left light source, front light source, right light source and rear light source, and only one adjustable wavelength light source lights up at a time.

[0077] Each time the adjustable wavelength light source is turned on, the wavelength of the adjustable wavelength light source is adjusted sequentially in order of increasing wavelength, and the industrial camera continuously records the image of each frame to obtain the recognition frame image;

[0078] Acquire the recognition frame image, perform grayscale processing on the recognition frame image, obtain the grayscale value of each pixel, and name it as single-point grayscale. When analyzing any pixel, name it as the point to be analyzed, name the grayscale value of the point to be analyzed as the grayscale value to be analyzed, name the pixels in the eight neighborhoods of the point to be analyzed as the neighboring points to be analyzed, name the grayscale values ​​of the neighboring points to be analyzed as the neighboring point grayscale values, calculate the absolute value of the difference between the grayscale value to be analyzed and the grayscale value of each neighboring point, name it as the difference value to be analyzed, obtain the maximum value among the differences to be analyzed, name it as the maximum difference value, calculate the maximum difference value of each pixel's low point, and then calculate the sum of all the maximum differences, name it as the grayscale difference degree.

[0079] The image frame with the largest grayscale difference is obtained and named the initial defect image.

[0080] In practical applications, when the light guide plate is fed into the defect scanning stage, the geometric center of the light guide plate overlaps with the geometric center of the bottom of the defect scanning stage, and at least one edge of the light guide plate is parallel to the left, front, right, or rear side. The light guide plate is illuminated sequentially in the order of left light source, front light source, right light source, and rear light source. This allows the shadow areas on the light guide plate in four directions to be collected. However, before extracting the defect area, the applicable wavelength needs to be determined. Therefore, the wavelengths of the adjustable wavelength light sources are adjusted sequentially in ascending order of wavelength. Each frame of the image is continuously recorded by an industrial camera to obtain the recognition frame image. Analyzing the grayscale difference is essentially analyzing the grayscale differences between adjacent pixels on the light guide plate. The greater the difference in grayscale value, the more obvious and easily observable the defect. Therefore, the identification frame image with the largest grayscale difference is used as the initial image of the defect. For example, the grayscale value of a certain pixel is 152, and the grayscale values ​​of its eight neighboring pixels include 138, 151, 149, 155, 156, 153, and 152. The difference values ​​to be analyzed are calculated to be 14, 1, 3, 3, 4, 1, and 0, respectively. The maximum difference value is further extracted to be 14. Each pixel has a maximum difference value. Each pixel is analyzed to obtain the grayscale difference value. It should be noted that when shooting the light guide plate, only the area where the light guide plate is located is shot. The background part is not included in any of the images.

[0081] The image preprocessing module is used to preprocess the defect recognition image to obtain the defect recognition image. Specifically, the image preprocessing of the defect recognition image involves image enhancement and grayscale processing to finally obtain the defect recognition image.

[0082] In practical applications, image enhancement improves the visual effect of an image by adjusting its contrast and brightness or using histogram equalization. Enhanced images can display details more clearly, improving readability and analytical performance. Image enhancement and grayscale processing are existing technologies, so they will not be described in detail in this embodiment.

[0083] The region identification module is used to perform defect analysis on the defect identification image, extract the defect boundary in the defect identification image, and obtain the defect region from the defect boundary; the region identification module includes a point to be confirmed analysis unit and a defect region analysis unit.

[0084] The unconfirmed point analysis unit is configured with an unconfirmed point analysis strategy, which includes:

[0085] The defect identification images captured under illumination from the left light source, front light source, right light source, and rear light source are named the left defect identification image, front defect identification image, right defect identification image, and rear defect identification image, respectively. When analyzing any defect identification image, it is named the target analysis image.

[0086] Please see Figure 2 As shown, the pixels in the target analysis map are named target image points, the gray values ​​of the target image points are named target image gray values, the target image point in the i-th row and j-th column of the target analysis map is marked as P(i,j), where i and j are both non-zero natural numbers and (i,j) is the index of P, and the target image gray value of P(i,j) is marked as H(i,j).

[0087] Label each row's P(i,j) as PH in top-to-bottom order, working by row. i In each column, P(i,j) is labeled as PL in order from left to right. j ;

[0088] In practical applications, since the analysis processes for the left defect identification image, the front defect identification image, the right defect identification image, and the rear defect identification image are basically the same, this embodiment only uses the left defect identification image as the target analysis image to illustrate the analysis process of this embodiment in detail; for example, a target analysis image like... Figure 2 As shown, its resolution is 9600×2400, meaning there are 9600 pixels per row and 2400 pixels per column. The original resolution of the industrial camera is 9600×6000. Figure 2 After removing the background portion, the width is narrowed, leaving a width of only 2400 pixels. The final labels are P(1,1) to P(2400,9600) and H(1,1) to H(2400,9600), and PH is also obtained. i and PL j , 1≤i≤2400, 1≤j≤9600.

[0089] If the target analysis map is a left defect identification map or a right defect identification map, then obtain PH. i At the same time, pH i The P(i,j) in the image are numbered and represented by the symbol TH(i,n), where n is a non-zero natural number and the range of n is the same as that of j. The smaller n is in TH(i,n), the closer the target image point is to the lit tunable wavelength light source.

[0090] If the target analysis diagram is a pre-defect identification diagram or a post-defect identification diagram, then obtain PL. j At the same time, for PL jThe P(i,j) in the image are numbered and represented by the symbol TL(j,m), where m is a non-zero natural number and the range of m is the same as that of i. The smaller m is in TL(j,m), the closer the target image point is to the lit tunable wavelength light source.

[0091] In practical applications, for example, in this embodiment, the target analysis map is the left defect identification map, then the PH is obtained. i The P(i,j) values ​​are numbered to obtain TH(i,n). The smaller the value of j, the closer it is to the left light source. Therefore, TH(i,1) to TH(i,9600) are P(i,1) to P(i,9600) respectively. If the target analysis map is a post-defect identification map, the larger the value of j, the closer it is to the right light source. In this case, TH(i,1) to TH(i,9600) are P(i,9600) to P(i,1) respectively. The same applies if the target analysis map is a pre-defect identification map or a post-defect identification map. This embodiment will not be described in detail.

[0092] Label TH(i,n) and TL(j,m) as TR(h,g), where h and g are both non-zero natural numbers and (h,g) is the index of TR. When the target analysis image is a left defect recognition image or a right defect recognition image, TR(h,g) represents TH(i,n). When the target analysis image is a front defect recognition image or a back defect recognition image, TR(h,g) represents TL(j,m). Label the grayscale of the target image corresponding to TR(h,g) as TE(h,g).

[0093] Please see Figure 3 As shown, each value of h is analyzed independently. Starting with g=2, |TE(h,g)-TE(h,g-1)| is calculated, and the result is marked as CL(h,g). g is incremented by one and CL(h,g) is calculated repeatedly until the maximum value of g is reached.

[0094] A one-dimensional coordinate system is established with CL(h,g) as the X-axis and named the preliminary defect analysis diagram. CL(h,g) is entered into the preliminary defect analysis diagram according to g. Cluster analysis is performed on the preliminary defect analysis diagram using a clustering algorithm to obtain difference clusters. The number of CL(h,g) in the difference clusters is counted and named the cluster size. The difference cluster with the largest cluster size is removed. The remaining difference clusters are collectively referred to as the defect to be confirmed clusters. TR(h,g) corresponding to CL(h,g) in the defect to be confirmed clusters are named the defect to be confirmed points.

[0095] In practical applications, since the target analysis image listed in this embodiment is the left defect identification image, TR(h,g) is equivalent to TH(i,n), and TR(h,1) to TR(h,9600) are equivalent to P(i,1) to P(i,9600), where h is i and g is n. TE(h,g) is also obtained by labeling, for example, the grayscale of some target images is as follows: Figure 3 As shown, Figure 3 The TE(h,g) values ​​in the image are arranged from left to right as TE(1,1) to TE(1,7). Each TE(h,g) has a grayscale value. Due to the small differences and lack of obvious contrast, they are difficult to distinguish with the naked eye, but can be directly extracted from the image. TE(1,1) to TE(1,7) are 133, 133, 132, 131, 132, 132, and 133 respectively. The calculated CL(1,2) to CL(1,7) are 0, 1, 1, 1, 0, and 1 respectively. Similarly, all CL(h,g) are calculated. Since the light guide plates are made of the same material and under the same lighting conditions, the normal surface of the light guide plate... The differences in grayscale values ​​are very small, with only the defective areas showing significant grayscale value changes. Therefore, a preliminary defect analysis map is constructed for cluster analysis. Since most areas on the light guide plate are normal areas with small grayscale value changes, the difference cluster with the largest size must belong to the normal surface CL(h,g). At the same time, since the grayscale value differences on the normal surface of the light guide plate are very small, these CL(h,g) can usually be classified into the same difference cluster. If errors occur occasionally, they can be removed when analyzing the defective areas later. Therefore, the remaining difference clusters are collectively referred to as defect confirmation clusters, and defect confirmation points are obtained.

[0096] The defect region analysis unit is used to further analyze the defect points to be confirmed, obtain the defect boundary, and obtain the defect region from the defect boundary;

[0097] The defect region analysis unit is configured with a defect region analysis strategy, which includes:

[0098] Analyze the defect points to be confirmed for each value of h. If any two defect points to be confirmed are adjacent in the target analysis diagram, they are integrated into a defect line to be confirmed. If any defect point to be confirmed is adjacent to any defect line to be confirmed, the defect point to be confirmed is included in the defect line to be confirmed.

[0099] Please see Figure 4 As shown, after the target analysis map is analyzed, different defect confirmation lines are obtained. The defect confirmation lines obtained from the analysis of the left defect identification map, the front defect identification map, the right defect identification map and the rear defect identification map are colored in the target analysis map. All target image points on the defect confirmation lines are filled with pure white to obtain the defect boundary.

[0100] Extract the closed region enclosed by the defect boundary and name it the defect region;

[0101] In practical applications, determining adjacency means determining whether two defect confirmation points are within each other's eight-neighbor domain. If so, these two defect confirmation points are adjacent. A scratch boundary typically has multiple defect confirmation points, and they must be consecutively adjacent, thus obtaining the defect confirmation line. The left, front, right, and rear defect identification images all contain corresponding defect confirmation lines. A scratch is essentially a groove, which must have boundaries in four directions; that is, the defect boundaries can enclose a closed area, and this closed area is the defect region. For example, the magnified defect boundary of a light guide plate... Figure 4 As shown, the defect boundary on the left forms a closed region, and the area enclosed by the white pixels on the left is the defect region. However, the defect boundary on the right does not form a closed region, so it can be determined that it is a false positive and does not constitute a defect region. It should be noted that... Figure 4 The enlarged defect boundary is represented by each pixel in the form of a square. The defect boundary is not the line between the white and gray parts, but rather the entire white part that forms the defect boundary line.

[0102] The perception analysis module is used to name the light guide plate with defective areas as the test board, and to perform perception tests on the test board to obtain the perception threshold through testing and analysis.

[0103] The perception and analysis module is configured with perception and analysis strategies, which include:

[0104] The light guide plate with defective areas is named the test plate. The test plate is used for light guiding. The tester observes the test plate and judges whether there is a defect. If the test plate is judged to have a defect, it is marked as a defective plate. If the test plate is judged not to have a defect, it is marked as a normal plate.

[0105] The defective area in the defective board is named the sensitive area, and the pixel in the sensitive area is named the sensitive point. The defective area in the normal board is named the non-sensitive area, and the pixel in the non-sensitive area is named the non-sensitive point.

[0106] In practical applications, light guide plates are widely used in advertising light boxes, LCD backlights, and home lighting, all of which assist human visual observation. Therefore, the final quality of a light guide plate depends on whether it affects visual observation. If the impact of defects cannot be observed by the naked eye during the use of the light guide plate, then the light guide plate is actually a qualified product. Since machines cannot completely simulate human eye recognition, the perception analysis module can only be executed manually. However, in this embodiment, only the initial testing phase requires manual participation. After the perception threshold is obtained through analysis, the manual testing phase can be canceled. In the initial manual testing phase, defective and normal plates are manually selected, and the sensitive areas, sensitive points, non-sensitive areas, and non-sensitive points are marked.

[0107] Get the difference of j between P(i,j) corresponding to the same row of the sensed points in the sensed region, named the sensed row width; get the difference of i between P(i,j) corresponding to the same column of the sensed points in the sensed region, named the sensed column width; get the maximum value of the sensed row width and sensed column width, named the single area width; get the minimum value of the single area width in the sensed regions of all defective boards, named the minimum sensed defect width.

[0108] In practical applications, with Figure 4 Taking the defective area in the middle as an example, Figure 4 The defect area has perceptible points in row 164, namely P(164,1346), P(164,1347), and P(164,1349). That is, within the pixels of row 164, the perceptible row width of the defect area is 1349-1346=3. Calculating the perceptible row width and perceptible column width only requires calculating the difference between the maximum and minimum values ​​of j and i, and 3 is only... Figure 4 The example defect area has a perceptible row width of 164 rows. Similarly, calculate all perceptible row widths and perceptible column widths of the defect area, and then obtain the maximum value to get the single area width. The single area width represents the number of pixels occupied at the widest point of this defect area. The larger the single area width, the easier it is to be observed by the naked eye. The perceptible area needs to refer to the minimum value of the single area width as a threshold to obtain the minimum perceptible defect width. The minimum perceptible defect width is the minimum value of the single area width that can be observed by the naked eye. If it is less than the minimum perceptible defect width, it cannot be observed by the naked eye, that is, the defect area has no impact on normal use, and vice versa.

[0109] Get the difference of j between P(i,j) corresponding to the same row of the non-sensitive points in the non-sensitive area, and name it as the non-sensitive row width. Get the difference of i between P(i,j) corresponding to the same column of the non-sensitive points in the non-sensitive area, and name it as the non-sensitive column width. Get the maximum value of the non-sensitive row width and non-sensitive column width of the non-sensitive area of ​​all defective boards, and name it as the maximum non-sensitive defect width.

[0110] The average of the minimum perceptible defect width and the maximum non-perceptible defect width is used to obtain the perception threshold.

[0111] In practical applications, the process of calculating the non-sensory row width and non-sensory column width is exactly the same as the process of calculating the sensory row width and sensory column width. Therefore, it will not be described in detail in this embodiment. Finally, the maximum value of the non-sensory row width and non-sensory column width is extracted to obtain the maximum non-sensory defect width. The maximum non-sensory defect width represents the maximum width of a single area that can be tolerated under the premise that it cannot be observed by the naked eye. Therefore, it is necessary to select the maximum value of the non-sensory row width and non-sensory column width. However, there is still a certain redundancy between the minimum sensory defect width and the maximum non-sensory defect width. Moreover, the perception test cannot analyze all situations. Therefore, their average value is calculated to obtain the perception threshold. Calculating the perception threshold is to introduce redundancy into the minimum sensory defect width, so that the minimum sensory defect width is more accurate.

[0112] The impact judgment module is used to extract the defect parameters of the defect area and refer to the perception threshold to determine whether the defects of the light guide plate have no impact.

[0113] The impact judgment module is configured with impact judgment strategies, which include:

[0114] Name the light guide plate that needs to be analyzed as the plate to be analyzed, and extract the defect area of ​​the plate to be analyzed and name it as the area to be analyzed.

[0115] Extract the width of a single region in the area to be analyzed and name it the defect parameter. Compare the defect parameter with the defect threshold. If the defect parameter is less than the defect threshold, output a surface defect signal; otherwise, output a surface qualified signal.

[0116] If the output surface fails to meet the requirements, it indicates that the light guide plate is defective, meaning the light guide plate is unqualified.

[0117] In practical applications, the process of extracting defect parameters of the area to be analyzed is exactly the same as the process of extracting the width of a single area of ​​the sensitive area. The criteria for judging the quality of the light guide plate have also been clearly given, and this embodiment will not be described in detail.

[0118] Example 2, please refer to Figure 5 As shown, this application provides a method for analyzing defects in a light guide plate, comprising the following steps:

[0119] Step S1 involves setting up a defect scanning stage, irradiating the light guide plate under different wavelengths, and capturing initial images of the defects. Step S1 includes the following sub-steps:

[0120] Step S101: Set up a defect scanning workbench and configure an adjustable wavelength light source and an industrial camera.

[0121] Step S101 includes the following sub-steps:

[0122] Step S101.1: Construct a closed, dark space, named the collection room, and name the side of the collection room the light source side. The light source side includes the left side, front side, right side, and rear side.

[0123] Step S101.2: Install an industrial camera at the geometric center of the top of the acquisition chamber;

[0124] Step S101.3: Connect the midpoints of the upper and lower edges of the light source surface to obtain the mounting line, install the slide rail along the mounting line, and install the adjustable wavelength light source on the slide rail.

[0125] Step S101.4: The adjustable wavelength light source is always aligned with the geometric center of the bottom of the acquisition chamber. The adjustable wavelength light source is moved in order from low to high. A flawed light guide plate is placed at the geometric center of the bottom of the acquisition chamber. The initial image of the defect is captured in real time. The initial image of the defect with the most obvious flaw is selected by the operator as the reference image. The position of the adjustable wavelength light source corresponding to the reference image is marked as the installation point. The adjustable wavelength light source is fixed at the installation point. The acquisition chamber, the adjustable wavelength light source and the industrial camera together form a defect scanning workbench.

[0126] Step S102: Acquire initial images and spectral diagrams of defects using the defect scanning workbench;

[0127] Step S102 includes the following sub-steps:

[0128] Step S102.1: The light guide plate is fed into the defect scanning stage, and the geometric center of the light guide plate overlaps with the geometric center of the bottom of the defect scanning stage.

[0129] Step S102.2: The adjustable wavelength light sources installed on the left side, front side, right side and rear side are named left light source, front light source, right light source and rear light source respectively. The light guide plate is illuminated in the order of left light source, front light source, right light source and rear light source, and only one adjustable wavelength light source lights up at a time.

[0130] Step S102.3: Each time the adjustable wavelength light source is turned on, the wavelength of the adjustable wavelength light source is adjusted sequentially in order from small to large wavelength, and the image of each frame is continuously recorded by the industrial camera to obtain the recognition frame image.

[0131] Step S102.4: Obtain the recognition frame image, perform grayscale processing on the recognition frame image, obtain the grayscale value of each pixel, and name it as single-point grayscale. When analyzing any pixel, name it as the point to be analyzed, name the grayscale value of the point to be analyzed as the grayscale value to be analyzed, name the pixels in the eight neighborhoods of the point to be analyzed as the neighboring points to be analyzed, name the grayscale values ​​of the neighboring points to be analyzed as the neighboring point grayscale values, calculate the absolute value of the difference between the grayscale value to be analyzed and the grayscale value of each neighboring point, name it as the difference value to be analyzed, obtain the maximum value among the differences to be analyzed, name it as the maximum difference value, calculate the maximum difference value of each pixel's low point, and then calculate the sum of all the maximum differences, name it as the grayscale difference degree.

[0132] Step S102.5: Obtain the recognition frame image with the largest grayscale difference and name it the initial defect image;

[0133] Step S2: Perform image preprocessing on the defect identification image to obtain the defect identification image; specifically, perform image enhancement and grayscale processing on the defect identification image to finally obtain the defect identification image.

[0134] Step S3 involves performing defect analysis on the defect recognition image, extracting the defect boundaries from the defect recognition image, and obtaining the defect region from the defect boundaries. Step S3 includes the following sub-steps:

[0135] Step S301: The defect recognition images captured when illuminated by the left light source, front light source, right light source and rear light source are named left defect recognition image, front defect recognition image, right defect recognition image and rear defect recognition image respectively. When analyzing any defect recognition image, it is named target analysis image.

[0136] Step S302: Name the pixels in the target analysis image as target image points, name the gray values ​​of the target image points as target image gray values, mark the target image points in the i-th row and j-th column of the target analysis image as P(i,j), where i and j are both non-zero natural numbers and (i,j) is the index of P, and mark the target image gray values ​​of P(i,j) as H(i,j);

[0137] Step S303: Label each row's P(i,j) as PH in a top-to-bottom order, row by row. i In each column, P(i,j) is labeled as PL in order from left to right. j ;

[0138] Step S304: If the target analysis map is a left defect identification map or a right defect identification map, then obtain PH. i At the same time, pH iThe P(i,j) in the image are numbered and represented by the symbol TH(i,n), where n is a non-zero natural number and the range of n is the same as that of j. The smaller n is in TH(i,n), the closer the target image point is to the lit tunable wavelength light source.

[0139] Step S305: If the target analysis map is a pre-defect identification map or a post-defect identification map, then obtain PL. j At the same time, for PL j The P(i,j) in the image are numbered and represented by the symbol TL(j,m), where m is a non-zero natural number and the range of m is the same as that of i. The smaller m is in TL(j,m), the closer the target image point is to the lit tunable wavelength light source.

[0140] Step S306: Mark TH(i,n) and TL(j,m) as TR(h,g), where h and g are both non-zero natural numbers and (h,g) is the index of TR. When the target analysis image is a left defect recognition image or a right defect recognition image, TR(h,g) represents TH(i,n). When the target analysis image is a front defect recognition image or a back defect recognition image, TR(h,g) represents TL(j,m). Mark the grayscale of the target image corresponding to TR(h,g) as TE(h,g).

[0141] Step S307: Analyze each value of h independently. Starting with g=2, calculate |TE(h,g)-TE(h,g-1)|, mark the calculation result as CL(h,g), increment g by one and repeat the calculation of CL(h,g) until the maximum value of g is reached.

[0142] Step S308: Establish a one-dimensional coordinate system with CL(h,g) as the X-axis and name it the preliminary defect analysis diagram. Input CL(h,g) into the preliminary defect analysis diagram according to g. Perform cluster analysis on the preliminary defect analysis diagram using a clustering algorithm to obtain difference clusters. Count the number of CL(h,g) in the difference clusters and name them as cluster size. Remove the difference cluster with the largest cluster size. Collect the remaining difference clusters as the defect to be confirmed clusters. Name the TR(h,g) corresponding to CL(h,g) in the defect to be confirmed clusters as defect to be confirmed points.

[0143] Step S309: Further analyze the defect points to be confirmed to obtain the defect boundary, and obtain the defect area from the defect boundary;

[0144] Step S309 includes the following sub-steps:

[0145] Step S309.1: Analyze the defect points to be confirmed for each value of h. If any two defect points to be confirmed are adjacent in the target analysis diagram, integrate them into a defect line to be confirmed. If any defect point to be confirmed is adjacent to any defect line to be confirmed, include the defect point to be confirmed in the defect line to be confirmed.

[0146] Step S309.2: After the target analysis map is analyzed, different defect confirmation lines are obtained. The defect confirmation lines obtained from the analysis of the left defect identification map, the front defect identification map, the right defect identification map and the rear defect identification map are colored in the target analysis map. All target image points on the defect confirmation lines are filled with pure white to obtain the defect boundary.

[0147] Step S309.3: Extract the closed region enclosed by the defect boundary and name it the defect region;

[0148] Step S4: The light guide plate with defective areas is named the test plate, and a perception test is performed on the test plate to obtain the perception threshold through testing and analysis; Step S4 includes the following sub-steps:

[0149] Step S401: The light guide plate with the defective area is named the test plate. The test plate is used for light guiding. The tester observes the test plate and judges whether there is a defect. If the test plate is judged to have a defect, it is marked as a defective plate. If the test plate is judged not to have a defect, it is marked as a normal plate.

[0150] Step S402: Name the defective area in the defective board as a sensitive area, name the pixels in the sensitive area as sensitive points, name the defective area in the normal board as a non-sensitive area, and name the pixels in the non-sensitive area as non-sensitive points.

[0151] Step S403: Obtain the difference of j between P(i,j) corresponding to the sensed points in the same row in the sensed region, and name it as sensed row width; obtain the difference of i between P(i,j) corresponding to the sensed points in the same column in the sensed region, and name it as sensed column width; obtain the maximum value between sensed row width and sensed column width, and name it as single area width; obtain the minimum value of single area width in the sensed regions of all defective boards, and name it as minimum sensed defect width.

[0152] Step S404: Obtain the difference of j between P(i,j) corresponding to the non-sensitive points in the same row in the non-sensitive area, and name it as the non-sensitive row width; obtain the difference of i between P(i,j) corresponding to the non-sensitive points in the same column in the non-sensitive area, and name it as the non-sensitive column width; obtain the maximum value of the non-sensitive row width and non-sensitive column width of the non-sensitive area of ​​all defective boards, and name it as the maximum non-sensitive defect width.

[0153] Step S405: Calculate the average of the minimum perceptible defect width and the maximum non-perceptible defect width to obtain the perception threshold;

[0154] Step S5: Extract the defect parameters of the defect area and determine whether the defect in the light guide plate has no impact by referring to the perception threshold; Step S5 includes the following sub-steps:

[0155] Step S501: Name the light guide plate that needs to be analyzed as the plate to be analyzed, extract the defect area of ​​the plate to be analyzed, and name it as the area to be analyzed.

[0156] Step S502: Extract the width of a single region of the area to be analyzed and name it as a defect parameter. Compare the defect parameter with the defect threshold. If the defect parameter is less than the defect threshold, output a surface defect signal; otherwise, output a surface qualified signal.

[0157] In step S503, if the output surface is unqualified, then the light guide plate is marked as defective, i.e., the light guide plate is unqualified.

[0158] Example 3: This application provides an electronic device, which may include a processor, a communication interface, a memory, and a communication bus. The processor, communication interface, and memory communicate with each other via the communication bus. The memory stores computer-readable instructions, and the processor can call these instructions. When the processor executes a computer-readable instruction, it performs steps as described in a light guide plate defect analysis method to achieve the following functions: 1. Construct a defect scanning workbench, irradiate the light guide plate under different wavelength conditions, and capture initial defect images; 2. Perform image preprocessing on the defect recognition images to obtain defect recognition images; 3. Perform defect analysis on the defect recognition images, extract defect boundaries from the defect recognition images, and obtain defect regions from the defect boundaries; 4. Name the light guide plate with defect regions as a test plate, perform a perception test on the test plate, and obtain a perception threshold through testing and analysis; 5. Extract defect parameters from the defect regions, and determine whether the defects in the light guide plate have no impact based on the perception threshold.

[0159] Furthermore, when the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0160] Example 4: This application also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it performs the steps of the above-described light guide plate defect analysis method to achieve the following functions: 1. Construct a defect scanning workbench, irradiate the light guide plate under different wavelength conditions, and capture initial defect images; 2. Perform image preprocessing on the defect recognition images to obtain defect recognition images; 3. Perform defect analysis on the defect recognition images, extract defect boundaries from the defect recognition images, and obtain defect regions from the defect boundaries; 4. Name the light guide plate with defect regions as a test plate, perform a perception test on the test plate, and obtain a perception threshold through testing and analysis; 5. Extract defect parameters from the defect regions, and determine whether the defects of the light guide plate have no impact by referring to the perception threshold.

[0161] Based on the above description of the embodiments, the embodiments of the present invention can be provided as methods, systems, or computer program products. Based on this understanding, the above technical solutions, in essence or in terms of their contribution to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or certain parts of the embodiments.

[0162] In the embodiments provided in this application, it should be understood that the disclosed system or method can be implemented in other ways. The embodiments described above are merely illustrative. For example, the division of modules or units is only a logical functional division, and there may be other division methods in actual implementation. Furthermore, multiple modules or units may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the coupling or direct coupling or communication connection shown or discussed may be through some communication interfaces. The indirect coupling or communication connection between systems, modules, and units may be electrical, mechanical, or other forms.

[0163] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for analyzing defects in a light guide plate, characterized in that, Includes the following steps: A defect scanning workbench was set up, and the light guide plate was irradiated under different wavelength conditions to obtain initial images of the defects. Image preprocessing is performed on the defect identification image to obtain the defect identification image; Defect analysis is performed on the defect recognition image to extract the defect boundary in the defect recognition image, and the defect region is obtained from the defect boundary; The light guide plate with defective areas is named the test plate, and a perception test is performed on the test plate to obtain the perception threshold through testing and analysis. Extract the defect parameters of the defect area and refer to the perception threshold to determine whether the defects of the light guide plate have no impact; Defect analysis is performed on the defect recognition image, the defect boundaries are extracted from the defect recognition image, and the defect region is obtained from the defect boundaries. This includes the following sub-steps: The defect identification images captured under illumination from the left light source, front light source, right light source, and rear light source are named the left defect identification image, front defect identification image, right defect identification image, and rear defect identification image, respectively. When analyzing any defect identification image, it is named the target analysis image. The pixels in the target analysis image are named target image points, the gray values ​​of the target image points are named target image gray values, the target image point in the i-th row and j-th column of the target analysis image is labeled as P(i,j), where i and j are both non-zero natural numbers and (i,j) is the index of P, and the target image gray value of P(i,j) is labeled as H(i,j). Label each row's P(i,j) as PH in top-to-bottom order, working by row. i In each column, P(i,j) is labeled as PL in order from left to right. j ; If the target analysis map is a left defect identification map or a right defect identification map, then obtain PH. i At the same time, pH i The P(i,j) in the figure are numbered and represented by the symbol TH(i,n), where n is a non-zero natural number and the value range of n is the same as that of j. The smaller n in TH(i,n), the closer the target image point is to the lit adjustable wavelength light source. If the target analysis diagram is a pre-defect identification diagram or a post-defect identification diagram, then obtain PL. j At the same time, for PL j The P(i,j) in the image are numbered and represented by the symbol TL(j,m), where m is a non-zero natural number and the range of m is the same as that of i. The smaller m is in TL(j,m), the closer the target image point is to the lit tunable wavelength light source. Label TH(i,n) and TL(j,m) as TR(h,g), where h and g are both non-zero natural numbers and (h,g) is the index of TR. When the target analysis image is a left defect recognition image or a right defect recognition image, TR(h,g) represents TH(i,n). When the target analysis image is a front defect recognition image or a back defect recognition image, TR(h,g) represents TL(j,m). Label the grayscale of the target image corresponding to TR(h,g) as TE(h,g). Analyze each value of h independently. Starting with g=2, calculate |TE(h,g)-TE(h,g-1)|, mark the calculation result as CL(h,g), increment g and repeat the calculation of CL(h,g) until the maximum value of g is reached. A one-dimensional coordinate system is established with CL(h,g) as the X-axis and named the preliminary defect analysis diagram. CL(h,g) is entered into the preliminary defect analysis diagram according to g. Cluster analysis is performed on the preliminary defect analysis diagram using a clustering algorithm to obtain difference clusters. The number of CL(h,g) in the difference clusters is counted and named the cluster size. The difference cluster with the largest cluster size is removed. The remaining difference clusters are collectively referred to as the defect to be confirmed clusters. TR(h,g) corresponding to CL(h,g) in the defect to be confirmed clusters are named the defect to be confirmed points. Further analysis is conducted on the defect points to be confirmed to obtain the defect boundary, and the defect area is obtained from the defect boundary; Further analysis of the defect points to be confirmed is performed to obtain the defect boundary, and the defect region is derived from the defect boundary, including the following sub-steps: Analyze the defect points to be confirmed for each value of h. If any two defect points to be confirmed are adjacent in the target analysis diagram, they are integrated into a defect line to be confirmed. If any defect point to be confirmed is adjacent to any defect line to be confirmed, the defect point to be confirmed is included in the defect line to be confirmed. After analyzing the target analysis map, different defect confirmation lines are obtained. The defect confirmation lines obtained from the analysis of the left defect identification map, the front defect identification map, the right defect identification map, and the rear defect identification map are colored in the target analysis map. All target image points on the defect confirmation lines are filled with pure white to obtain the defect boundaries. Extract the closed region enclosed by the defect boundary and name it the defect region; The light guide plate with defective areas is named the test plate, and a perception test is performed on the test plate. The perception threshold is obtained by testing and analysis, including the following sub-steps: The light guide plate with defective areas is named the test plate. The test plate is used for light guiding. The tester observes the test plate and judges whether there is a defect. If the test plate is judged to have a defect, it is marked as a defective plate. If the test plate is judged not to have a defect, it is marked as a normal plate. The defective area in the defective board is named the sensitive area, and the pixel in the sensitive area is named the sensitive point. The defective area in the normal board is named the non-sensitive area, and the pixel in the non-sensitive area is named the non-sensitive point. Get the difference of j between P(i,j) corresponding to the same row of the sensed points in the sensed region, named the sensed row width; get the difference of i between P(i,j) corresponding to the same column of the sensed points in the sensed region, named the sensed column width; get the maximum value of the sensed row width and sensed column width, named the single area width; get the minimum value of the single area width in the sensed regions of all defective boards, named the minimum sensed defect width. Get the difference of j between P(i,j) corresponding to the same row of the non-sensitive points in the non-sensitive area, and name it as the non-sensitive row width. Get the difference of i between P(i,j) corresponding to the same column of the non-sensitive points in the non-sensitive area, and name it as the non-sensitive column width. Get the maximum value of the non-sensitive row width and non-sensitive column width of the non-sensitive area of ​​all defective boards, and name it as the maximum non-sensitive defect width. The average of the minimum perceptible defect width and the maximum non-perceptible defect width is used to obtain the perception threshold.

2. The method for analyzing defects in a light guide plate according to claim 1, characterized in that, The process of setting up a defect scanning stage, irradiating the light guide plate under different wavelengths, and capturing initial images of the defects includes the following sub-steps: Set up a defect scanning workbench and equip it with an adjustable wavelength light source and an industrial camera; Initial images and spectral diagrams of defects are acquired using a defect scanning workbench.

3. The method for analyzing defects in a light guide plate according to claim 2, characterized in that, Setting up a defect scanning workbench and configuring an adjustable wavelength light source and industrial camera includes the following sub-steps: Construct a closed, dark space and name it the collection room. Name the side of the collection room the light source side, which includes the left side, front side, right side, and rear side. An industrial camera is installed at the geometric center of the top of the acquisition chamber; Connect the midpoints of the upper and lower edges of the light source surface to obtain the mounting line. Install the slide rail along the mounting line and mount the adjustable wavelength light source on the slide rail. The adjustable wavelength light source is always aligned with the geometric center of the bottom of the acquisition chamber. The adjustable wavelength light source is moved in ascending order, and a flawed light guide plate is placed at the geometric center of the bottom of the acquisition chamber to capture initial images of defects in real time. The initial image of the defect with the most obvious flaw is manually selected as the reference image. The position of the adjustable wavelength light source corresponding to the reference image is marked as the installation point, and the adjustable wavelength light source is fixed at the installation point. The acquisition chamber, the adjustable wavelength light source, and the industrial camera together form a defect scanning workbench.

4. The method for analyzing defects in a light guide plate according to claim 3, characterized in that, Acquiring initial images and spectral maps of defects using a defect scanning workbench includes the following sub-steps: The light guide plate is fed into the defect scanning stage, and the geometric center of the light guide plate overlaps with the geometric center of the bottom of the defect scanning stage. The adjustable wavelength light sources installed on the left side, front side, right side and rear side are named left light source, front light source, right light source and rear light source respectively. The light guide plate is illuminated in the order of left light source, front light source, right light source and rear light source, and only one adjustable wavelength light source lights up at a time. Each time the adjustable wavelength light source is turned on, the wavelength of the adjustable wavelength light source is adjusted sequentially in order of increasing wavelength, and the industrial camera continuously records the image of each frame to obtain the recognition frame image; Acquire the recognition frame image, perform grayscale processing on the recognition frame image, obtain the grayscale value of each pixel, and name it as single-point grayscale. When analyzing any pixel, name it as the point to be analyzed, name the grayscale value of the point to be analyzed as the grayscale value to be analyzed, name the pixels in the eight neighborhoods of the point to be analyzed as the neighboring points to be analyzed, name the grayscale values ​​of the neighboring points to be analyzed as the neighboring point grayscale values, calculate the absolute value of the difference between the grayscale value to be analyzed and the grayscale value of each neighboring point, name it as the difference value to be analyzed, obtain the maximum value among the differences to be analyzed, name it as the maximum difference value, calculate the maximum difference value of each pixel's low point, and then calculate the sum of all the maximum differences, name it as the grayscale difference degree. Obtain the recognition frame image with the largest grayscale difference and name it the initial defect image.

5. The method for analyzing defects in a light guide plate according to claim 4, characterized in that, Image preprocessing for defect identification images specifically involves image enhancement and grayscale conversion, ultimately yielding the defect identification image.

6. The method for analyzing defects in a light guide plate according to claim 5, characterized in that, Extracting defect parameters from the defective area and determining whether the defects in the light guide plate have no impact, based on a reference perception threshold, includes the following sub-steps: Name the light guide plate that needs to be defect-analyzed as the plate to be analyzed, and extract the defect area of ​​the plate to be analyzed and name it as the area to be analyzed. Extract the width of a single region in the area to be analyzed and name it the defect parameter. Compare the defect parameter with the sensing threshold. If the defect parameter is greater than the sensing threshold, output a surface defect signal; otherwise, output a surface qualified signal. If the output surface fails to meet the requirements, it indicates that the light guide plate is defective, meaning the light guide plate is unqualified.

7. A light guide plate defect analysis system, used to implement the light guide plate defect analysis method according to any one of claims 1-6, characterized in that, It includes a defect acquisition module, an image preprocessing module, a region recognition module, a perception analysis module, and an impact judgment module; the defect acquisition module, the image preprocessing module, the perception analysis module, and the impact judgment module are all data connected to the region recognition module. The defect acquisition module is used to build a defect scanning workbench, irradiate the light guide plate under different wavelength conditions, and capture initial images of the defects. The image preprocessing module is used to preprocess the defect identification image to obtain the defect identification image; The region recognition module is used to perform defect analysis on the defect recognition image, extract the defect boundary in the defect recognition image, and obtain the defect region from the defect boundary; The perception analysis module is used to name the light guide plate with defective areas as the test plate, and to perform perception tests on the test plate to obtain the perception threshold through testing and analysis. The impact judgment module is used to extract the defect parameters of the defect area and determine whether the defects of the light guide plate have no impact by referring to the perception threshold.

Images