A light guide plate hot-pressing line pre-AOI detection system and detection method

Abstract

The application discloses a light guide plate hot-pressing line pre-AOI detection system and a detection method, relates to the technical field of light guide plate detection, and comprises the following steps: judging whether defects exist and acquiring abnormal profiles based on real-time whiteboard binaryzation images; constructing a real-time detection ratio based on the abnormal profiles; acquiring a second number of light guide plate hot-pressing pre-graphs with scratch defects, and marking the graphs as historical scratch graphs; acquiring a detection ratio threshold based on the historical scratch graphs; and dividing first-type defects and second-type defects based on the real-time detection ratio and the detection ratio threshold. The application is used to solve the problem that a simple defect type division is lacked in the existing light guide plate detection technology, thereby affecting the production efficiency of subsequent light guide plates.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of light guide plate detection, in particular to a light guide plate hot-pressing line pre-AOI detection system and method. BACKGROUND

[0002] As the core component of the backlight module, the function of the light guide plate is to convert the linear LED light source into a uniformly distributed surface light source, which directly determines the brightness uniformity, luminance and visual effect of the liquid crystal display screen. At present, the mainstream high-quality light guide plate manufacturing generally adopts a hot-pressing process. This process forms microstructure optical dots on the surface of the optical-grade plate through precise molds under high temperature and high pressure, thereby controlling the refraction and scattering of light. However, the hot-pressing process itself has the characteristics of irreversibility and high cost. Once the defective substrate enters the hot-pressing line, any small surface scratch or foreign matter attached will be permanently solidified or even enlarged under high temperature and high pressure, resulting in waste products. This not only causes waste of valuable raw materials, but also causes huge losses such as equipment working hours, energy consumption and mold loss. Therefore, it is necessary to detect the AOI before the light guide plate hot-pressing line.

[0003] However, the defects identified by the existing technology can be divided into two categories. The first category is the defect of the light guide plate itself, such as scratches. The second category is the presence of foreign matter on the light guide plate. Because different defect types can perform different recycling operations on the light guide plate, such as the light guide plate with foreign matter, the light guide plate can continue to be used after removing the foreign matter. Therefore, in order to facilitate the subsequent recycling of the light guide plate, the defects of the light guide plate can be further divided into two categories. The existing technology lacks a simple defect type division, such as the patent application with the publication number CN109064451A, which discloses a light guide plate defect detection method. This scheme only identifies defects and does not further classify defect types, which affects the production efficiency of the subsequent light guide plate. Therefore, the existing light guide plate detection technology lacks a simple defect type division, which affects the production efficiency of the subsequent light guide plate. SUMMARY

[0004] The present application aims to at least solve one of the technical problems in the prior art. By judging whether there is a defect based on the real-time whiteboard binary image and obtaining an abnormal contour, constructing a real-time detection ratio based on the abnormal contour, obtaining a second number of light guide plates with scratch defects before hot-pressing, and marking them as historical scratch images, obtaining a detection ratio threshold based on the historical scratch images, and dividing the first and second categories of defects based on the real-time detection ratio and the detection ratio threshold, the present application solves the problem of lacking a simple defect type division in the existing light guide plate detection technology, which affects the production efficiency of the subsequent light guide plate.

[0005] To achieve the above-mentioned purpose, the present application provides a light guide plate hot-pressing line pre-AOI detection system, which comprises:

[0006] The image acquisition module, the gray processing module, the whiteboard threshold acquisition module, the binarization module, the contour acquisition module, the detection value construction module, the detection threshold acquisition module, and the type division module;

[0007] The image acquisition module is configured to acquire an image of a light guide plate before heat pressing, which is marked as a real-time whiteboard image;

[0008] The gray processing module is configured to perform gray processing on the real-time whiteboard image to obtain a real-time whiteboard gray image;

[0009] The whiteboard threshold acquisition module is configured to acquire a first number of images of light guide plates before heat pressing without defects, which are marked as historical whiteboard images; and acquire a whiteboard threshold based on the historical whiteboard images;

[0010] The binarization module is configured to perform binarization processing on the real-time whiteboard gray image based on the whiteboard threshold to obtain a real-time whiteboard binary image;

[0011] The contour acquisition module is configured to determine whether the real-time whiteboard binary image contains defects and acquire an abnormal contour based on the real-time whiteboard binary image;

[0012] The detection value construction module is configured to construct a real-time detection ratio based on the abnormal contour;

[0013] The detection threshold acquisition module is configured to acquire a second number of images of light guide plates before heat pressing with scratch defects, which are marked as historical scratch images; and acquire a detection ratio threshold based on the historical scratch images;

[0014] The type division module is configured to divide the first type of defects and the second type of defects based on the real-time detection ratio and the detection ratio threshold.

[0015] Further, the gray processing module is configured with a gray processing strategy, and the gray processing strategy includes:

[0016] The RGB value of each pixel point in the real-time whiteboard image is converted into a gray value using a gray conversion formula to obtain a real-time whiteboard gray image; and the gray value of the pixel point in the real-time whiteboard gray image is marked as a real-time whiteboard gray value.

[0017] Further, the whiteboard threshold acquisition module is configured with a whiteboard threshold acquisition strategy, and the whiteboard threshold acquisition strategy includes:

[0018] The historical whiteboard image is processed to obtain a historical whiteboard gray image; and the gray value of each pixel point in the historical whiteboard gray image is acquired and marked as a historical whiteboard gray value;

[0019] The range of the historical whiteboard gray value is acquired; and the range of the historical whiteboard gray value is divided into a third number of equal intervals, which are marked as real-time interval ranges;

[0020] Mark the minimum value in any one real-time interval range as the first interval value;

[0021] Get the number of historical whiteboard gray values in each real-time interval range, marked as real-time distribution number;

[0022] Sort the real-time distribution number from left to right according to the corresponding first interval value from small to large;

[0023] Calculate the first distribution threshold value as: F1=a×(Gs÷D3); Wherein F1 is the first distribution threshold value, a is a real number between 0 and 1, Gz is the sum of all real-time distribution numbers, and D3 is the third number;

[0024] Mark the real-time distribution number less than or equal to the first distribution threshold value as the first distribution abnormal number;

[0025] Determine whether the leftmost real-time distribution number is the first distribution abnormal number, if so, delete the real-time distribution number until the leftmost real-time distribution number is not the first distribution abnormal number; Get the first interval value corresponding to the leftmost real-time distribution number after deletion, marked as whiteboard threshold value.

[0026] Further, the binarization module is configured with a binarization strategy, and the binarization strategy comprises:

[0027] Set the real-time whiteboard gray value less than the whiteboard threshold value in the real-time whiteboard gray image to 0, and set the real-time whiteboard gray value greater than or equal to the whiteboard threshold value to 255, and obtain a real-time whiteboard binarization image.

[0028] Further, the contour acquisition module is configured with a contour acquisition strategy, and the contour acquisition strategy comprises:

[0029] Mark the pixel point with a gray value of 0 in the real-time whiteboard binarization image as an abnormal pixel point;

[0030] Get the independent area composed of abnormal pixel points, marked as initial abnormal area;

[0031] If no initial abnormal area appears, it is determined that the light guide plate to be detected has no defects; If the initial abnormal area appears, it is determined that the light guide plate to be detected has defects;

[0032] Get any one initial abnormal area adjacent to the abnormal pixel point with a gray value of 255, marked as boundary pixel point;

[0033] Get the contour composed of boundary pixel points, marked as abnormal contour.

[0034] Further, the detection value construction module is configured with a starting intersection point acquisition strategy, which comprises:

[0035] A plane rectangular coordinate system is established and marked as an image analysis coordinate system; the real-time whiteboard binary image is placed in the image analysis coordinate system;

[0036] A fourth number of coordinate points are drawn on the abnormal contour at equal intervals, and marked as image analysis coordinate points;

[0037] The image analysis coordinate points are sequentially connected along the abnormal contour to obtain line segments, which are marked as connecting line segments;

[0038] The angle between every two adjacent connecting line segments is marked as a line segment angle;

[0039] The intersection point of the two adjacent connecting line segments corresponding to the smallest line segment angle is obtained, which is marked as a starting intersection point.

[0040] Further, the detection value construction module is configured with a real-time detection ratio acquisition strategy, which comprises:

[0041] A movable coordinate point is constructed on the abnormal contour, which is marked as a moving coordinate point;

[0042] The starting intersection point and the moving coordinate point are connected to obtain a connecting line segment; the moving coordinate point is moved around the abnormal contour for one revolution, and the connecting line segment with the maximum length is obtained, which is marked as a first connecting line segment;

[0043] A perpendicular line of the first target line segment is drawn through the midpoint of the first target line segment, which is marked as a target perpendicular line;

[0044] A line segment connecting the two intersection points of the target perpendicular line and the abnormal contour is obtained, which is marked as a second connecting line segment;

[0045] The ratio of the lengths of the first connecting line segment and the second connecting line segment is calculated, which is marked as a real-time detection ratio.

[0046] Further, the detection threshold acquisition module is configured with a detection threshold acquisition strategy, which comprises:

[0047] The real-time detection ratio of each abnormal contour in the historical scratch map is obtained, which is marked as a historical detection ratio;

[0048] The range of the historical detection ratio is obtained; the range of the historical detection ratio is divided into a fifth number of equal interval ranges, which are marked as historical interval ranges;

[0049] The minimum value in any one historical interval range is marked as a second interval value;

[0050] Obtaining the number of historical detection ratios in each historical interval range, marked as historical distribution number;

[0051] Sorting the historical distribution number from small to large according to the corresponding second interval value from left to right;

[0052] The second distribution threshold is calculated as: F2=a2×(Gl÷D5); Wherein F2 is the first distribution threshold, a2 is a real number between 0 and 1, Gl is the sum of all historical distribution numbers, and D5 is the fifth number;

[0053] Marking the historical distribution number less than or equal to the second distribution threshold as the second distribution abnormal number;

[0054] Determine whether the leftmost historical distribution number is the second distribution abnormal number, if yes, delete the historical distribution number until the leftmost historical distribution number is not the second distribution abnormal number; Obtain the second interval value corresponding to the leftmost historical distribution number after deletion, marked as detection ratio threshold.

[0055] Further, the type division module is configured with a type division strategy, and the type division strategy comprises:

[0056] Determine whether the real-time detection ratio is greater than or equal to the detection ratio threshold, if yes, mark the abnormal contour as the first kind of defect, if not, mark the abnormal contour as the second kind of defect.

[0057] The application provides a light guide plate hot pressing line front AOI detection method, comprising the following steps: obtaining the image of the light guide plate before hot pressing, marked as real-time whiteboard image;

[0058] The real-time whiteboard image is subjected to gray scale processing to obtain a real-time whiteboard gray scale image;

[0059] Obtaining the first number of images of the light guide plate without defects before hot pressing, marked as historical whiteboard image; Obtaining the whiteboard threshold based on the historical whiteboard image;

[0060] Based on the whiteboard threshold, the real-time whiteboard gray scale image is subjected to binaryzation processing to obtain a real-time whiteboard binaryzation image;

[0061] Based on the real-time whiteboard binaryzation image, it is determined whether there is a defect and an abnormal contour is obtained;

[0062] Based on the abnormal contour, a real-time detection ratio is constructed;

[0063] Obtaining the second number of images of the light guide plate with scratch defects before hot pressing, marked as historical scratch image; Obtaining the detection ratio threshold based on the historical scratch image;

[0064] Based on the real-time detection ratio and the detection ratio threshold, the first kind of defect and the second kind of defect are divided.

[0065] The present application has the advantages that the present application can determine whether defects exist and obtain abnormal contours based on real-time whiteboard binary images; construct real-time detection ratios based on the abnormal contours; obtain a second number of images of light guide plates before hot pressing, which are marked as historical scratch images; obtain detection ratio thresholds based on the historical scratch images; and divide the first and second types of defects based on the real-time detection ratios and the detection ratio thresholds, which can preliminarily distinguish the scratches of the light guide plates from other types of defects, saves the processing time of the defective light guide plates, and improves the overall efficiency of producing the light guide plates.

[0066] The present application has the advantages that the present application can determine whether defects exist and obtain abnormal contours based on real-time whiteboard binary images; construct real-time detection ratios based on the abnormal contours; obtain a second number of images of light guide plates before hot pressing, which are marked as historical scratch images; obtain detection ratio thresholds based on the historical scratch images; and divide the first and second types of defects based on the real-time detection ratios and the detection ratio thresholds, which can preliminarily distinguish the scratches of the light guide plates from other types of defects, saves the processing time of the defective light guide plates, and improves the overall efficiency of producing the light guide plates. BRIEF DESCRIPTION OF DRAWINGS

[0067] Figure 1 The figure is a schematic diagram of the starting intersection of the present application;

[0068] Figure 2 The figure is a schematic diagram of the starting intersection of the present application;

[0069] Figure 3 The figure is a schematic diagram of the first and second connection line segments of the present application;

[0070] Figure 4 The figure is a flowchart of the steps of the present application. DETAILED DESCRIPTION

[0071] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0072] Embodiment 1, please refer to Figure 1 The figure is a schematic diagram of the starting intersection of the present application;

[0073] The image acquisition module is configured to acquire an image of the light guide plate before heat pressing, which is marked as a real-time whiteboard image; when the image is acquired, the light guide plate is placed in a ring-shaped light source, so that the light source is uniformly on the light guide plate to be detected, so that the defect part is more obviously displayed; at the same time, the light guide plate before heat pressing is a whiteboard, so the defect is more easily identified.

[0074] The gray processing module is configured to perform gray processing on the real-time whiteboard image to obtain a real-time whiteboard gray image.

[0075] The gray processing module is configured with a gray processing strategy, and the gray processing strategy includes:

[0076] The RGB value of each pixel point in the real-time whiteboard image is converted into a gray value by using a gray conversion formula to obtain a real-time whiteboard gray image; the gray value of the pixel point in the real-time whiteboard gray image is marked as a real-time whiteboard gray value; the gray conversion formula here is a weighted average conversion formula, that is, the mean value of R, G and B values is calculated.

[0077] The whiteboard threshold acquisition module is configured to acquire a first number of images of the light guide plate before heat pressing without defects, which is marked as a historical whiteboard image; the whiteboard threshold is acquired based on the historical whiteboard image; in the case of the same lighting condition, the first number is set to better acquire the gray value distribution of the light guide plate without defects, for example, the first number is 10.

[0078] The whiteboard threshold acquisition module is configured with a whiteboard threshold acquisition strategy, and the whiteboard threshold acquisition strategy includes:

[0079] The historical whiteboard image is processed to obtain a historical whiteboard gray image; the gray value of each pixel point in the historical whiteboard gray image is acquired and marked as a historical whiteboard gray value.

[0080] The range of the historical whiteboard gray value is acquired; the historical whiteboard gray value range is divided into a third number of equal intervals, which is marked as a real-time interval range; the third number of real-time interval ranges is set to better observe the distribution of the historical whiteboard gray value; the third number is not suitable to be too large or too small, for example, the third number is 10;

[0081] The minimum value in any one real-time interval range is marked as a first interval value;

[0082] The number of historical whiteboard gray values in each real-time interval range is acquired and marked as a real-time distribution number;

[0083] The real-time distribution number is sorted from left to right according to the corresponding first interval value from small to large;

[0084] The first distribution threshold is calculated as: F1=a×(Gs÷D3); wherein F1 is the first distribution threshold, a is a real number between 0 and 1, Gz is the sum of all real-time distribution numbers, and D3 is the third number; wherein Gs÷D3 is the average number of historical whiteboard grayscale values in each real-time interval range, the first distribution threshold is used to obtain a real-time interval range with a smaller real-time distribution number, and therefore the value of a should not be too large, for example, a is 0.01;

[0085] In actual application, for example, the sum of real-time distribution numbers is 2073 million, the first distribution threshold is calculated as: F1=0.01×(2073÷10)=2, and the calculation result is rounded to an integer, so the first distribution threshold is 2.

[0086] The real-time distribution number less than or equal to the first distribution threshold is marked as the first distribution abnormal number;

[0087] It is judged whether the leftmost real-time distribution number is the first distribution abnormal number, if yes, the real-time distribution number is deleted until the leftmost real-time distribution number is not the first distribution abnormal number; the first interval value corresponding to the leftmost real-time distribution number after deletion is obtained and marked as the whiteboard threshold;

[0088] In actual application, for example, the range of historical whiteboard grayscale values is 190 to 230; the range of historical whiteboard grayscale values is divided into a third number of equal interval ranges, which are 190 to 194, 194 to 198,..., 226 to 230 respectively; the real-time distribution numbers are 2, 23,..., 32, and 12 respectively; the real-time distribution number less than or equal to 2 is marked as the first distribution abnormal number, and the first distribution abnormal number includes 2; 2 is deleted, and the real-time distribution numbers are 23,..., 32, and 12 respectively; the first interval value corresponding to the real-time distribution number corresponding to 23 is obtained as 194, and the whiteboard threshold is 194; by this method, the historical whiteboard grayscale values that are abnormally small can be screened out.

[0089] The binarization module is configured to perform binarization processing on the real-time whiteboard grayscale image based on the whiteboard threshold to obtain a real-time whiteboard binary image;

[0090] The binarization module is configured with a binarization strategy, and the binarization strategy includes:

[0091] The real-time whiteboard grayscale value less than the whiteboard threshold in the real-time whiteboard grayscale image is set to 0, and the real-time whiteboard grayscale value greater than or equal to the whiteboard threshold is set to 255, and a real-time whiteboard binary image is obtained; because the whiteboard has high brightness under light, the real-time whiteboard grayscale value is large; if a defect occurs, that is, a shadow, the real-time whiteboard grayscale value is small; at the same time, the obtained image does not appear other interference images, and therefore the part greater than or equal to the whiteboard threshold can be regarded as a normal light guide plate part, and the part less than the whiteboard threshold can be regarded as a defect part.

[0092] In practical application, the real-time whiteboard grayscale value less than 194 in the real-time whiteboard grayscale image is set to 0, and the real-time whiteboard grayscale value greater than or equal to 194 is set to 255, to obtain a real-time whiteboard binary image.

[0093] The contour obtaining module is configured to determine whether the real-time whiteboard binary image contains defects and to obtain an abnormal contour based on the real-time whiteboard binary image.

[0094] The contour obtaining module is configured with a contour obtaining strategy, and the contour obtaining strategy includes:

[0095] The pixel point with a grayscale value of 0 in the real-time whiteboard binary image is marked as an abnormal pixel point.

[0096] An independent area composed of the abnormal pixel points is obtained and marked as an initial abnormal area; since the real-time whiteboard grayscale value is not within the grayscale value distribution range of a normal light guide plate, it can be preliminarily determined as a defect area.

[0097] If the initial abnormal area does not appear, it is determined that the light guide plate to be detected does not have defects; if the initial abnormal area appears, it is determined that the light guide plate to be detected has defects.

[0098] An abnormal pixel point adjacent to the grayscale value of 255 in any one initial abnormal area is obtained and marked as a boundary pixel point.

[0099] A contour composed of the boundary pixel points is obtained and marked as an abnormal contour.

[0100] The detection value construction module is configured to construct a real-time detection ratio based on the abnormal contour.

[0101] The detection value construction module is configured with a starting intersection point obtaining strategy, and the starting intersection point obtaining strategy includes:

[0102] A plane rectangular coordinate system is established and marked as an image analysis coordinate system; the real-time whiteboard binary image is placed in the image analysis coordinate system.

[0103] A fourth number of coordinate points are drawn on the abnormal contour at equal intervals, and marked as image analysis coordinate points.

[0104] The image analysis coordinate points are sequentially connected along the abnormal contour to obtain line segments, and the line segments are marked as connection line segments.

[0105] The angle between every two adjacent connection line segments is marked as a line segment angle.

[0106] The intersection point of the two adjacent connecting line segments corresponding to the minimum line segment angle is marked as a starting intersection point; the defect mainly identified here is a scratch, which is usually an elongated image. The abnormal profile of the elongated image is approximately a long strip, and the line segment angles on the abnormal profile are different in value, especially at the two ends of the long strip. The line segment angle at the two ends of the long strip, and the starting intersection point is approximately one end of the long strip-shaped abnormal profile. This method can quickly and accurately obtain one end of the long strip-shaped abnormal profile, saving time and operation complexity.

[0107] In practical applications, please refer to Figure 2 The line segment angle diagram obtained is shown.

[0108] The detection value construction module is configured with a real-time detection ratio acquisition strategy, and the real-time detection ratio acquisition strategy includes:

[0109] A movable coordinate point is constructed on the abnormal profile, which is marked as a moving coordinate point;

[0110] The starting intersection point and the moving coordinate point are connected to obtain a connecting line segment. The moving coordinate point is moved around the abnormal profile, and the connecting line segment with the maximum length is obtained, which is marked as a first connecting line segment. The first line segment can be approximately regarded as the length of the scratch;

[0111] A perpendicular line of the first target line segment passing through the midpoint of the first target line segment is drawn, which is marked as a target perpendicular line;

[0112] The line segment connecting the two intersection points of the target perpendicular line and the abnormal profile is obtained, which is marked as a second connecting line segment;

[0113] The ratio of the lengths of the first connecting line segment and the second connecting line segment is calculated, which is marked as a real-time detection ratio;

[0114] The beneficial effect of setting the real-time detection ratio is that the first line segment can be approximately regarded as the length of the scratch; the first line segment can be approximately regarded as the width of the scratch, i.e. the real-time detection ratio is the length-width ratio of the scratch; the scratch is elongated, so the real-time detection ratio is usually increased, and the rest of the defects or the attached objects on the light guide plate are mostly smaller than the real-time detection ratio of the scratch; therefore, the type of the scratch can be preliminarily judged by the real-time detection ratio.

[0115] In practical applications, please refer to Figure 3 The schematic diagram of the first connecting line segment and the second connecting line segment obtained is shown, the real-time detection ratio is 31, and the calculation result is rounded to an integer.

[0116] The detection threshold acquisition module is configured to obtain a second number of images of the light guide plate before hot pressing with the scratch defect, marked as a historical scratch image; obtain a detection ratio threshold based on the historical scratch image; and set the second number of historical scratch images to obtain a distribution range of the real-time detection ratio of the scratch; therefore, the second number should not be too small, for example, 100;

[0117] The detection threshold acquisition module is configured with a detection threshold acquisition strategy, and the detection threshold acquisition strategy includes:

[0118] Obtain the real-time detection ratio of each abnormal profile in the historical scratch image, marked as a historical detection ratio; the method for obtaining the historical detection ratio and the real-time detection ratio is the same;

[0119] Obtain the range of the historical detection ratio; divide the range of the historical detection ratio into a fifth number of equal intervals, marked as a historical interval range; set the fifth number of historical interval ranges to better observe the distribution of the historical detection ratio; therefore, the fifth number should not be too large or too small, for example, the fifth number is 10;

[0120] Mark the minimum value in any one historical interval range as a second interval value;

[0121] Obtain the number of historical detection ratios in each historical interval range, marked as a historical distribution number;

[0122] Sort the historical distribution numbers from left to right according to the corresponding second interval values from small to large;

[0123] Calculate the second distribution threshold as: F2=a2×(Gl÷D5); wherein F2 is the first distribution threshold, a2 is a real number between 0 and 1, Gl is the sum of all historical distribution numbers, and D5 is the fifth number; wherein Gl÷D5 is the average number of historical detection ratios in each historical interval range, and the second distribution threshold is to obtain the historical interval range with a smaller historical distribution number, therefore, a2 should not be too large, for example, a2 is 0.1;

[0124] In actual application, the sum of all historical distribution numbers is 112, and the second distribution threshold is calculated as: F2=0.1×(112÷10)=1; the calculation result is rounded to an integer, and therefore the second distribution threshold is 1.

[0125] Mark the historical distribution numbers less than or equal to the second distribution threshold as a second distribution abnormal number;

[0126] determining whether the leftmost historical distribution number is the second distribution abnormal number, if yes, deleting the historical distribution number until the leftmost historical distribution number is not the second distribution abnormal number; obtaining the second interval value corresponding to the leftmost historical distribution number after deletion, and marking it as a detection ratio threshold value; through this method, the real-time detection ratio value that is too small can be screened out;

[0127] In practical applications, for example, the range of historical detection ratio values is 6 to 36; the range of historical detection ratio values is divided into a fifth number of equal intervals, respectively 6 to 9, 9 to 12,..., 34 to 36; the real-time distribution number is 1, 8,..., 9, respectively; the historical distribution number less than or equal to 1 is marked as the second distribution abnormal number, and the first distribution abnormal number is 1. After deleting 1, the historical distribution number is 8,..., 9, respectively. The second interval value corresponding to the historical distribution number corresponding to 8 is 9, and the detection ratio threshold value is 9.

[0128] The type division module is configured to divide the first type of defects and the second type of defects based on the real-time detection ratio value and the detection ratio threshold value.

[0129] The type division module is configured with a type division strategy, and the type division strategy includes:

[0130] determining whether the real-time detection ratio value is greater than or equal to the detection ratio threshold value, if yes, marking the abnormal contour as the first type of defect, if not, marking the abnormal contour as the second type of defect; the detection ratio threshold value is identified as the minimum value of the real-time detection ratio value, that is, the minimum value of the real-time detection ratio value of the scratch, so that whether it is a scratch and the remaining defects can be preliminarily analyzed based on the real-time detection ratio value and the detection ratio threshold value, facilitating subsequent identification of the defect light guide plate.

[0131] In practical applications, the real-time detection ratio value is 31, which is greater than the detection ratio threshold value 9, so the abnormal contour is marked as the first type of defect, which can be preliminarily identified as a scratch.

[0132] Embodiment 2, please refer to Figure 4 As shown in the figure, a light guide plate hot-pressing line pre-AOI detection method includes the following steps:

[0133] Step S1, obtaining an image of the light guide plate before hot pressing for detection, and marking it as a real-time whiteboard image.

[0134] Step S2, performing grayscale processing on the real-time whiteboard image to obtain a real-time whiteboard grayscale image; Step S2 includes the following substeps:

[0135] Step S201, converting the RGB value of each pixel point in the real-time whiteboard image into a grayscale value using a grayscale conversion formula to obtain a real-time whiteboard grayscale image; the grayscale value of the pixel point in the real-time whiteboard grayscale image is marked as a real-time whiteboard grayscale value.

[0136] Step S3, obtaining a first number of images of the light guide plate before hot pressing without containing defects, marked as a historical whiteboard image; obtaining a whiteboard threshold based on the historical whiteboard image; step S3 includes the following sub-steps:

[0137] Step S301, obtaining a historical whiteboard grayscale image by performing grayscale processing on the historical whiteboard image; obtaining the grayscale value of each pixel point in the historical whiteboard grayscale image, marked as a historical whiteboard grayscale value;

[0138] Step S302, obtaining the range of the historical whiteboard grayscale value; dividing the range of the historical whiteboard grayscale value into a third number of equal intervals, marked as a real-time interval range;

[0139] Step S303, marking the minimum value in any one real-time interval range as a first interval value;

[0140] Step S304, obtaining the number of historical whiteboard grayscale values in each real-time interval range, marked as a real-time distribution number;

[0141] Step S305, sorting the real-time distribution numbers from left to right according to the corresponding first interval values from small to large;

[0142] Step S306, calculating the first distribution threshold as: F1=a×(Gs÷D3); wherein F1 is the first distribution threshold, a is a real number between 0 and 1, Gz is the sum of all real-time distribution numbers, and D3 is the third number;

[0143] Step S307, marking the real-time distribution numbers less than or equal to the first distribution threshold as a first distribution abnormal number;

[0144] Step S308, determining whether the leftmost real-time distribution number is the first distribution abnormal number, if so, deleting the real-time distribution number until the leftmost real-time distribution number is not the first distribution abnormal number; obtaining the first interval value corresponding to the leftmost real-time distribution number after deletion, marked as a whiteboard threshold.

[0145] Step S4, performing binaryzation processing on the real-time whiteboard grayscale image based on the whiteboard threshold to obtain a real-time whiteboard binaryzation image; step S4 includes the following sub-steps:

[0146] Step S401, setting the real-time whiteboard grayscale value less than the whiteboard threshold in the real-time whiteboard grayscale image to 0, and setting the real-time whiteboard grayscale value greater than or equal to the whiteboard threshold to 255, to obtain a real-time whiteboard binaryzation image.

[0147] Step S5, determining whether there is a defect based on the real-time whiteboard binaryzation image and obtaining an abnormal contour; step S5 includes the following sub-steps:

[0148] Step S501, mark the pixel points with a gray value of 0 in the real-time whiteboard binarization image as abnormal pixel points;

[0149] Step S502, obtain an independent region composed of abnormal pixel points and mark it as an initial abnormal region;

[0150] Step S503, if no initial abnormal region appears, determine that the light guide plate to be detected has no defects; if an initial abnormal region appears, determine that the light guide plate to be detected has defects;

[0151] Step S504, obtain an abnormal pixel point adjacent to a gray value of 255 in any one initial abnormal region and mark it as a boundary pixel point;

[0152] Step S505, obtain a contour composed of boundary pixel points and mark it as an abnormal contour.

[0153] Step S6, construct a real-time detection ratio based on the abnormal contour; step S6 includes the following sub-steps:

[0154] Step S601, establish a plane rectangular coordinate system and mark it as an image analysis coordinate system; place the real-time whiteboard binarization image in the image analysis coordinate system;

[0155] Step S602, draw fourth-numbered coordinate points on the abnormal contour at equal intervals and mark them as image analysis coordinate points;

[0156] Step S603, sequentially connect the image analysis coordinate points along the abnormal contour to obtain line segments and mark them as connecting line segments;

[0157] Step S604, mark the angle between every two adjacent connecting line segments as a line segment angle;

[0158] Step S605, obtain the intersection point of the two adjacent connecting line segments corresponding to the smallest line segment angle and mark it as a starting intersection point;

[0159] Step S606, construct a movable coordinate point on the abnormal contour and mark it as a moving coordinate point;

[0160] Step S607, connect the starting intersection point and the moving coordinate point to obtain a connecting line segment; move the moving coordinate point around the abnormal contour for one round to obtain the connecting line segment with the maximum length and mark it as a first connecting line segment;

[0161] Step S608, draw a perpendicular line of the first target line segment through the midpoint of the first target line segment and mark it as a target perpendicular line;

[0162] Step S609, obtain a line segment connecting the two intersection points of the target perpendicular line and the abnormal contour and mark it as a second connecting line segment;

[0163] Step S610, calculate the ratio of the length of the first connecting line segment and the second connecting line segment, and mark it as a real-time detection ratio.

[0164] Step S7, obtain a second number of images of the light guide plate before hot pressing with scratch defects, and mark it as a historical scratch image; obtain a detection ratio threshold based on the historical scratch image; step S7 includes the following sub-steps:

[0165] Step S701, obtain the real-time detection ratio of each abnormal profile in the historical scratch image, and mark it as a historical detection ratio;

[0166] Step S702, obtain the range of the historical detection ratio; divide the range of the historical detection ratio into a fifth number of equal intervals, and mark it as a historical interval range;

[0167] Step S703, mark the minimum value in any one historical interval range as a second interval value;

[0168] Step S704, obtain the number of historical detection ratios in each historical interval range, and mark it as a historical distribution number;

[0169] Step S705, sort the historical distribution numbers from left to right according to the corresponding second interval values from small to large;

[0170] Step S706, calculate the second distribution threshold as: F2=a2×(Gl÷D5); wherein F2 is the first distribution threshold, a2 is a real number between 0 and 1, Gl is the sum of all historical distribution numbers, and D5 is the fifth number;

[0171] Step S707, mark the historical distribution number less than or equal to the second distribution threshold as a second distribution abnormal number;

[0172] Step S708, determine whether the leftmost historical distribution number is the second distribution abnormal number, if so, delete the historical distribution number until the leftmost historical distribution number is not the second distribution abnormal number; obtain the second interval value corresponding to the leftmost historical distribution number after deletion, and mark it as a detection ratio threshold.

[0173] Step S8, divide the first kind of defects and the second kind of defects based on the real-time detection ratio and the detection ratio threshold; step S8 includes the following sub-steps:

[0174] Step S801, determine whether the real-time detection ratio is greater than or equal to the detection ratio threshold, if so, mark the abnormal profile as the first kind of defect, if not, mark the abnormal profile as the second kind of defect.

[0175] Those skilled in the art will appreciate that embodiments of the present application can be readily used as a method, a system or a computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (or computer- readable storage media) having computer-usable program code embodied in the medium. The medium can be any available storage media that can be accessed by a computer. By way of example, and not limitation, such computer-usable storage media can include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other storage medium that can be used to carry or store desired computer program code in the form of instructions or data structures and that can be accessed by a computer. Also, the present application can be embodied in a computer program product that can be traded as goods or merchandise, through the storage medium described above or any other suitable medium. When the computer program code is executed by a computer, a series of instructions or a series of operations is performed. Also, the present application can be embodied in a computer program product that can be traded as goods or merchandise, through the storage medium described above or any other suitable medium. When the computer program code is executed by a computer, a series of instructions or a series of operations is performed. Figure 1 one or more flows and / or blocks Figure 1 one or more flows and / or blocks

[0176] In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented in other manners. The embodiments described above are merely specific implementation manners of the present application, and the partition shown or discussed as above between the units is merely logical function division, and there can be another partition manner in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed coupling or direct coupling or communication connection between the units can be indirect coupling or communication connection through some interface, and can be electrical, mechanical or other forms.

Claims

1. A light guide plate hot line pre-AOI detection system, characterized in that, The application relates to a light guide plate defect detection method and device. The method comprises the following steps: An image acquisition module, a gray processing module, a whiteboard threshold acquisition module, a binary module, a contour acquisition module, a detection value construction module, a detection threshold acquisition module and a type division module are included. The image acquisition module is used for acquiring an image of a light guide plate before heat pressing, which is marked as a real-time whiteboard image. The gray processing module is used for carrying out gray processing on the real-time whiteboard image to obtain a real-time whiteboard gray image. The whiteboard threshold acquisition module is used for acquiring a first number of images of light guide plates without defects before heat pressing, which are marked as historical whiteboard images; and a whiteboard threshold is acquired based on the historical whiteboard images. The binary module is used for carrying out binary processing on the real-time whiteboard gray image based on the whiteboard threshold to obtain a real-time whiteboard binary image. The contour acquisition module is used for judging whether the real-time whiteboard binary image contains defects and acquiring an abnormal contour. The detection value construction module is used for constructing a real-time detection ratio based on the abnormal contour. The detection threshold acquisition module is used for acquiring a second number of images of light guide plates with scratch defects before heat pressing, which are marked as historical scratch images; and a detection ratio threshold is acquired based on the historical scratch images. The type division module is used for dividing the first type of defects and the second type of defects based on the real-time detection ratio and the detection ratio threshold. The detection value construction module is configured with a real-time detection ratio acquisition strategy, and the real-time detection ratio acquisition strategy comprises the following steps: An abnormal contour is constructed on the abnormal contour, which is marked as a moving coordinate point. A connecting line segment is obtained by connecting the starting intersection point and the moving coordinate point; the moving coordinate point is moved around the abnormal contour for one round, and the connecting line segment with the maximum length is obtained, which is marked as a first connecting line segment. A vertical line of the first target line segment is drawn through the midpoint of the first target line segment, which is marked as a target vertical line. A line segment connected by two intersection points of the target vertical line and the abnormal contour is obtained, which is marked as a second connecting line segment.

2. The system according to claim 1, wherein, The ratio of the lengths of the first connecting line segment and the second connecting line segment is calculated, which is marked as a real-time detection ratio. The gray processing module is configured with a gray processing strategy, and the gray processing strategy comprises the following steps:

3. The system according to claim 2, wherein the system further comprises a light source and a camera. The RGB value of each pixel point in the real-time whiteboard image is converted into a gray value by using a gray conversion formula to obtain a real-time whiteboard gray image; and the gray value of the pixel point in the real-time whiteboard gray image is marked as a real-time whiteboard gray value. The whiteboard threshold acquisition module is configured with a whiteboard threshold acquisition strategy, and the whiteboard threshold acquisition strategy comprises the following steps: The historical whiteboard image is processed to obtain a historical whiteboard gray image; and the gray value of each pixel point in the historical whiteboard gray image is obtained and marked as a historical whiteboard gray value. The range of the historical whiteboard gray value is obtained; the range is divided into a third number of equal intervals, which are marked as real-time interval ranges. The minimum value in any one real-time interval range is marked as a first interval value. The number of the historical whiteboard gray values in each real-time interval range is obtained, which is marked as a real-time distribution number. The real-time distribution numbers are sorted from left to right according to the corresponding first interval values from small to large. The first distribution threshold is calculated as: F1=a×(Gs÷D3); wherein F1 is the first distribution threshold, a is a real number between 0 and 1, Gs is the sum of all real-time distribution numbers, and D3 is the third number; The real-time distribution number less than or equal to the first distribution threshold is marked as the first distribution abnormal number; It is judged whether the leftmost real-time distribution number is the first distribution abnormal number, if yes, the real-time distribution number is deleted until the leftmost real-time distribution number is not the first distribution abnormal number; the first interval value corresponding to the leftmost real-time distribution number after deletion is obtained and marked as the whiteboard threshold.

4. The system according to claim 3, wherein the system further comprises a light source and a camera. The binarization module is configured with a binarization strategy, and the binarization strategy comprises: The real-time whiteboard grayscale value less than the whiteboard threshold in the real-time whiteboard grayscale image is set to 0, and the real-time whiteboard grayscale value greater than or equal to the whiteboard threshold is set to 255, and a real-time whiteboard binarization image is obtained.

5. The system according to claim 4, wherein the system further comprises a light source and a camera. The contour acquisition module is configured with a contour acquisition strategy, and the contour acquisition strategy comprises: The pixel point with a grayscale value of 0 in the real-time whiteboard binarization image is marked as an abnormal pixel point; An independent area composed of the abnormal pixel points is obtained and marked as an initial abnormal area; If the initial abnormal area does not appear, it is determined that the light guide plate to be detected has no defects; if the initial abnormal area appears, it is determined that the light guide plate to be detected has defects; An arbitrary abnormal pixel point adjacent to the grayscale value of 255 in any one initial abnormal area is marked as a boundary pixel point; An outline composed of the boundary pixel points is obtained and marked as an abnormal outline.

6. The system according to claim 5, wherein the system further comprises a light source and a camera. The detection value construction module is configured with a starting intersection point acquisition strategy, and the starting intersection point acquisition strategy comprises: A plane rectangular coordinate system is established and marked as an image analysis coordinate system; the real-time whiteboard binarization image is placed in the image analysis coordinate system; A fourth number of coordinate points are drawn on the abnormal outline at equal intervals, and marked as image analysis coordinate points; The image analysis coordinate points are sequentially connected along the abnormal outline to obtain line segments, and marked as connected line segments; The included angle between every two adjacent connected line segments is marked as a line segment included angle; The intersection point of the two adjacent connected line segments corresponding to the minimum line segment included angle is obtained and marked as a starting intersection point.

7. The system according to claim 6, wherein the system further comprises a light source and a camera. The detection threshold acquisition module is configured with a detection threshold acquisition strategy, and the detection threshold acquisition strategy comprises: The real-time detection ratio of each abnormal outline in the historical scratch image is obtained and marked as a historical detection ratio; The range of the historical detection ratio is obtained; the range of the historical detection ratio is divided into a fifth number of equal interval ranges, and marked as historical interval ranges; The minimum value in any one historical interval range is marked as a second interval value; The number of historical detection ratios in each historical interval range is obtained and marked as a historical distribution number; The historical distribution numbers are sorted from left to right according to the corresponding second interval values from small to large; The second distribution threshold is calculated as: F2=a2×(Gl÷D5); wherein F2 is the second distribution threshold, a2 is a real number between 0 and 1, Gl is the sum of all historical distribution numbers, and D5 is the fifth number; The historical distribution number less than or equal to the second distribution threshold is marked as a second distribution abnormal number; It is judged whether the leftmost historical distribution number is the second distribution abnormal number, and if yes, the historical distribution number is deleted until the leftmost historical distribution number is not the second distribution abnormal number; a second interval value corresponding to the leftmost historical distribution number after deletion is obtained and marked as a detection ratio threshold.

8. The system according to claim 7, wherein the system further comprises a light source and a camera. The type classification module is configured with a type classification strategy, and the type classification strategy comprises: It is judged whether the real-time detection ratio value is greater than or equal to the detection ratio threshold, and if yes, the abnormal profile is marked as the first type of defect, and if not, the abnormal profile is marked as the second type of defect.

9. A light guide plate hot line before AOI detection method, suitable for the light guide plate hot line before AOI detection system of any one of claims 1-8, characterized in that, The method comprises the following steps: An image of the light guide plate before heat pressing is obtained and marked as a real-time whiteboard image; The real-time whiteboard image is subjected to grayscale processing to obtain a real-time whiteboard grayscale image; A first number of images of the light guide plate without defects before heat pressing are obtained and marked as historical whiteboard images; a whiteboard threshold is obtained based on the historical whiteboard images; The real-time whiteboard grayscale image is subjected to binary processing based on the whiteboard threshold to obtain a real-time whiteboard binary image; It is judged whether the real-time whiteboard binary image contains defects and an abnormal profile is obtained; A real-time detection ratio value is constructed based on the abnormal profile; A second number of images of the light guide plate with scratch defects before heat pressing are obtained and marked as historical scratch images; a detection ratio threshold is obtained based on the historical scratch images; The first type of defect and the second type of defect are divided based on the real-time detection ratio value and the detection ratio threshold.

Images