Display screen defect detection method and system based on machine vision, and medium

Abstract

The application provides a display screen defect detection method and system based on machine vision and a medium. The method comprises the following steps: setting light source parameters to form a detection lighting environment, and placing a display screen panel in the lighting environment; collecting an image of the display screen panel under a high gray scale background based on a high dynamic industrial camera to obtain a display screen panel image; preprocessing the display screen panel image to obtain a pretreated image; extracting image features based on the pretreated image, and screening out spot features and texture features in the image based on the image features; extracting defect features based on a threshold segmentation algorithm for the spot features and the texture features to obtain defect information; classifying the defect features based on the defect information to obtain different defect types, and generating a detection report; and dynamically adjusting light source parameters of the lighting environment to provide a better detection environment, segmenting the defect features based on the threshold segmentation algorithm, accurately analyzing spot defects of the display screen panel, and improving detection accuracy.

Description

Technical Field

[0001] This application relates to the field of display screen defect detection technology, and more specifically, to a display screen defect detection method, system, and medium based on machine vision. Background Technology

[0002] Mura defects are common uneven brightness and color defects that occur during the production of flat panel displays, and they manifest in the following ways:

[0003] Brightness Mura: Local area brightness deviation (high brightness or low brightness);

[0004] Color Mura: Uneven distribution of R / G / B sub-pixel regions causes color shift;

[0005] Complex distribution patterns such as blocky Mura, striped Mura, and cloud-like Mura;

[0006] It has a large area but very little grayscale variation, making it easily overlooked.

[0007] The challenge of Mura's flaws lies in:

[0008] Low visual contrast, typically ΔL ≤ 5%; (ΔL refers to Luminance Difference, which is an indicator that measures the degree of brightness unevenness between a local area on the display screen and its surrounding background, and is one of the core quantitative parameters for detecting Mura (spot-type brightness unevenness defects).

[0009] The shapes vary greatly, making it difficult to template them; traditional edge detection or bright spot recognition methods cannot be used; a balance between detection sensitivity and false detection rate must be ensured.

[0010] Currently, mainstream LCD manufacturers use the following testing methods:

[0011] Manual visual inspection is inefficient and highly subjective;

[0012] Single-threshold image subtraction has poor sensitivity and weak anti-interference ability;

[0013] Spatial frequency domain analysis requires the support of a high-precision uniform light source and imaging system;

[0014] AI classification and detection models have poor generalization ability and lack interpretability. Summary of the Invention

[0015] The purpose of this application is to provide a machine vision-based method, system, and medium for detecting defects in a display screen. By dynamically adjusting the light source parameters of the lighting environment, a better detection environment is provided. The defect features are segmented according to a threshold segmentation algorithm, and the spot defects of the display screen panel are accurately identified and classified, thereby reducing the false detection rate and environmental sensitivity and improving the detection accuracy.

[0016] This application also provides a machine vision-based method for detecting defects in a display screen, including:

[0017] Set the light source parameters to create the lighting environment for detection, and place the display panel in the lighting environment;

[0018] The image of the display panel is obtained by acquiring the image of the display panel against a high grayscale background using a high dynamic range industrial camera. The display panel image is then preprocessed to obtain a preprocessed image.

[0019] Image features are extracted from the preprocessed image, and blob and texture features are selected from the image based on these features.

[0020] Defect features are extracted from spot and texture features using a threshold segmentation algorithm to obtain defect information.

[0021] Based on the defect information, the defect features are classified to obtain different defect types, and an inspection report is generated.

[0022] Optionally, in the machine vision-based display screen defect detection method described in this application embodiment, setting light source parameters to form a detection lighting environment and placing the display screen panel in the lighting environment specifically includes:

[0023] Set the light source parameters, which include light source brightness, light source uniformity, and color temperature value;

[0024] Multiple light source parameter combinations are matched based on the ratio of light source brightness, light source uniformity, and color temperature value;

[0025] Multiple lighting environments are formed based on combinations of light source parameters, and the presence of local bright spots in these environments is analyzed.

[0026] If present, adjust the ratio of light source brightness, light source uniformity, and color temperature value;

[0027] If there are no local bright spots, the required lighting environment for testing is obtained. Place the display screen in the center of the lighting environment and adjust the angle and level of the display screen.

[0028] Optionally, in the machine vision-based display screen defect detection method described in this application embodiment, obtaining the display screen panel image by acquiring an image of the display screen panel against a high grayscale background using a high dynamic range industrial camera specifically includes:

[0029] Acquire the original display panel image, and perform grayscale processing on the original display panel image to obtain a grayscale image;

[0030] Calculate the gray value of each pixel in the grayscale image, and average the gray values ​​of all pixels to obtain the average gray value of the current image.

[0031] Set the target desired gray value, calculate the difference between the current image's average gray value and the target desired gray value, and obtain the gray value difference;

[0032] The light source intensity is analyzed based on the lighting environment, the camera exposure time is obtained, and the light source intensity and exposure time are dynamically adjusted based on the grayscale difference to obtain the final display panel image.

[0033] Optionally, in the machine vision-based display screen defect detection method described in this application embodiment, the display screen panel image is preprocessed to obtain a preprocessed image, specifically including:

[0034] Obtain the final display panel image and mark the pixel coordinates of the four corner points on the display panel image;

[0035] A transformation matrix is ​​constructed based on the four corner calibration points. The ideal coordinates of the corresponding points in the target image are set. The corner pixel coordinates are transformed to the ideal coordinates based on the transformation matrix to obtain the corrected image.

[0036] The corrected image is then normalized to obtain a fixed-size image;

[0037] Based on a fixed-size image, the brightness values ​​of different pixels in the image are analyzed, and the brightness uniformity is analyzed based on the brightness values ​​of all pixels.

[0038] The brightness uniformity is compared with the set uniformity threshold to obtain the uniformity difference. Brightness value compensation is then performed based on the uniformity difference to obtain the preprocessed image.

[0039] Optionally, in the machine vision-based display screen defect detection method described in this application embodiment, defect feature extraction is performed on spot features and texture features based on a threshold segmentation algorithm to obtain defect information, specifically including:

[0040] Obtain the preprocessed image, and perform grayscale normalization on the preprocessed image to obtain the normalized image;

[0041] Extract normalized image features, set a first segmentation threshold, and segment the normalized image features based on the segmentation threshold to obtain image foreground features and image background features.

[0042] A second segmentation threshold is set, and the foreground features of the image are segmented based on the second segmentation threshold to obtain blob features and texture features;

[0043] A defect threshold is set, and defect feature segmentation is performed on the spot features and texture features based on the defect threshold to obtain defect information.

[0044] Optionally, in the machine vision-based display screen defect detection method described in this application embodiment, the defect features are classified based on defect information to obtain different defect types, and a detection report is generated, specifically including:

[0045] Obtain several defect samples including different defect categories, and divide the defect samples into training set and test set according to a set ratio;

[0046] Build an initial model framework, and iteratively train the initial model framework based on the training set to obtain the training results;

[0047] Determine whether the training results have converged;

[0048] If convergence is achieved, the model is tested on the test set, and the hyperparameters of the model are dynamically adjusted according to the test results to obtain the defect classification model. Based on the defect classification model, the defect information is analyzed to obtain different defect types.

[0049] Different defect types are classified and labeled to generate inspection reports;

[0050] If the model does not converge, adjust the number of iterations until it converges.

[0051] Secondly, embodiments of this application provide a machine vision-based display screen defect detection system. The system includes a memory and a processor. The memory includes a program for a machine vision-based display screen defect detection method. When executed by the processor, the program for the machine vision-based display screen defect detection method performs the following steps:

[0052] Set the light source parameters to create the lighting environment for detection, and place the display panel in the lighting environment;

[0053] The image of the display panel is obtained by acquiring the image of the display panel against a high grayscale background using a high dynamic range industrial camera. The display panel image is then preprocessed to obtain a preprocessed image.

[0054] Image features are extracted from the preprocessed image, and blob and texture features are selected from the image based on these features.

[0055] Defect features are extracted from spot and texture features using a threshold segmentation algorithm to obtain defect information.

[0056] Based on the defect information, the defect features are classified to obtain different defect types, and an inspection report is generated.

[0057] Optionally, in the machine vision-based display screen defect detection system described in this application embodiment, setting light source parameters to form a detection lighting environment and placing the display screen panel in the lighting environment specifically includes:

[0058] Set the light source parameters, which include light source brightness, light source uniformity, and color temperature value;

[0059] Multiple light source parameter combinations are matched based on the ratio of light source brightness, light source uniformity, and color temperature value;

[0060] Multiple lighting environments are formed based on combinations of light source parameters, and the presence of local bright spots in these environments is analyzed.

[0061] If present, adjust the ratio of light source brightness, light source uniformity, and color temperature value;

[0062] If there are no local bright spots, the required lighting environment for testing is obtained. Place the display screen in the center of the lighting environment and adjust the angle and level of the display screen.

[0063] Optionally, in the machine vision-based display defect detection system described in this application embodiment, the display panel image is obtained by acquiring an image of the display panel against a high grayscale background using a high dynamic range industrial camera, specifically including:

[0064] Acquire the original display panel image, and perform grayscale processing on the original display panel image to obtain a grayscale image;

[0065] Calculate the gray value of each pixel in the grayscale image, and average the gray values ​​of all pixels to obtain the average gray value of the current image.

[0066] Set the target desired gray value, calculate the difference between the current image's average gray value and the target desired gray value, and obtain the gray value difference;

[0067] The light source intensity is analyzed based on the lighting environment, the camera exposure time is obtained, and the light source intensity and exposure time are dynamically adjusted based on the grayscale difference to obtain the final display panel image.

[0068] Thirdly, embodiments of this application also provide a computer-readable storage medium, which includes a machine vision-based display screen defect detection method program. When the machine vision-based display screen defect detection method program is executed by a processor, it implements the steps of the machine vision-based display screen defect detection method as described in any of the preceding claims.

[0069] As can be seen from the above, the display screen defect detection method, system, and medium provided in this application embodiment, based on machine vision, forms a detection lighting environment by setting light source parameters and placing the display screen panel in the lighting environment; images of the display screen panel against a high grayscale background are acquired using a high dynamic range industrial camera to obtain display screen panel images; the display screen panel images are preprocessed to obtain preprocessed images; image features are extracted based on the preprocessed images, and spot features and texture features are filtered out from the images based on the image features; defect features are extracted from the spot features and texture features based on a threshold segmentation algorithm to obtain defect information; the defect features are classified based on the defect information to obtain different defect types, and a detection report is generated; by dynamically adjusting the light source parameters of the lighting environment to provide a better detection environment, and by segmenting the defect features according to the threshold segmentation algorithm, the spot defects of the display screen panel are accurately analyzed, accurately identified, and classified, reducing the false detection rate and environmental sensitivity, and improving detection accuracy. Attached Figure Description

[0070] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0071] Figure 1 A flowchart illustrating a machine vision-based display screen defect detection method provided in an embodiment of this application;

[0072] Figure 2 A flowchart illustrating the lighting environment setup for a machine vision-based display defect detection method provided in this application embodiment;

[0073] Figure 3 A flowchart illustrating the process of acquiring display panel images in a machine vision-based display defect detection method provided in this application embodiment;

[0074] Figure 4 A block diagram of a machine vision-based display screen defect detection system provided in an embodiment of this application. Detailed Implementation

[0075] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0076] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0077] Please refer to Figure 1 , Figure 1 This is a flowchart of a machine vision-based display screen defect detection method according to some embodiments of this application. This machine vision-based display screen defect detection method is used in a terminal device and includes the following steps:

[0078] S101, Set the light source parameters to create a detection lighting environment, and place the display panel in the lighting environment;

[0079] S102: Based on a high dynamic range industrial camera, an image of the display panel is acquired against a high grayscale background to obtain the display panel image. The display panel image is then preprocessed to obtain a preprocessed image.

[0080] S103, extract image features based on preprocessed images, and filter out blob features and texture features in the image based on image features;

[0081] S104, Defect features are extracted from spot features and texture features based on threshold segmentation algorithm to obtain defect information;

[0082] S105: Based on the defect information, the defect features are classified to obtain different defect types, and an inspection report is generated.

[0083] It should be noted that extracting blob and texture features from an image involves three stages:

[0084] (1) Gray-scale normalization:

[0085] formula:

[0086] ,

[0087] meaning: : Original image pixels;

[0088] The local average gray level with radius rrr centered on the pixel;

[0089] Normalize the image to eliminate brightness gradient interference.

[0090] (2) Spatial frequency enhancement:

[0091] The edge response is calculated using the Sobel gradient operator:

[0092] ,

[0093] SFR(x, y) represents the Spatial Frequency Response at point (x, y) in the image.

[0094] Inorm is used in the image normalization process, and its specific formula is as follows:

[0095] ,

[0096] The meaning is explained as follows:

[0097] Original image at points The pixel grayscale value;

[0098] :by Centered on, with radius as The average gray value of a local area;

[0099] Inorm: The normalized image grayscale value, used to suppress the interference of background brightness gradient changes on defect detection;

[0100] This normalization operation can effectively eliminate large-scale brightness unevenness, making low-contrast defects (such as Mura) more obvious in the background, thus improving the accuracy of subsequent feature extraction and defect segmentation.

[0101] Symbol explanation:

[0102] : Gray-level gradient in the x-direction;

[0103] : Gray-scale gradient in the y-direction.

[0104] (3) Fourier domain high-frequency filtering:

[0105] Frequency domain enhancement operation:

[0106] ,

[0107] This represents the response value of the high-pass filter at coordinates (u, v) in the two-dimensional frequency space.

[0108] Fourier transform / inverse transform;

[0109] High-pass filters, typically Butterworth high-pass filters:

[0110] ,

[0111] These represent coordinate points in the frequency domain, corresponding to the frequency components in the horizontal and vertical directions of the image, respectively. They are the frequency indices of the filter's location.

[0112] This represents the center coordinates of the spectrum, i.e., the location of the DC component (usually at the center of the image after Fourier transform). This is the reference point for calculating frequency distances; This indicates the cutoff frequency; n indicates the filter order.

[0113] Please refer to Figure 2 , Figure 2 This is a flowchart illustrating the lighting environment setup for a machine vision-based display screen defect detection method according to some embodiments of this application. According to embodiments of the present invention, setting light source parameters to form a detection lighting environment and placing the display screen panel within this lighting environment specifically includes:

[0114] S201, Set the light source parameters, which include light source brightness, light source uniformity, and color temperature value;

[0115] S202, which is based on the ratio of light source brightness, light source uniformity and color temperature value to multiple light source parameter combinations;

[0116] S203, based on the combination of light source parameters, multiple lighting environments are formed, and the presence of local bright spots in the lighting environments is analyzed;

[0117] S204, if present, adjust the ratio of light source brightness, light source uniformity, and color temperature value;

[0118] S205. If there are no local bright spots, the required lighting environment for testing is obtained. The display screen is placed in the center of the lighting environment, and the angle and level of the display screen are adjusted.

[0119] It should be noted that by setting different proportions of light source brightness, light source uniformity, and color temperature, multiple combinations of light source parameters are obtained. By analyzing whether the lighting environment formed by these multiple light source parameter combinations meets the requirements, the proportions of light source brightness, light source uniformity, and color temperature can be precisely adjusted to obtain the optimal lighting environment.

[0120] Please refer to Figure 3 , Figure 3 This is a flowchart illustrating the image acquisition process of a display panel using a machine vision-based display defect detection method according to some embodiments of this application. According to an embodiment of the present invention, an image of the display panel is obtained by acquiring an image of the display panel against a high grayscale background using a high dynamic range industrial camera, specifically including:

[0121] S301, acquire the original display panel image, perform grayscale processing on the original display panel image to obtain a grayscale image;

[0122] S302, calculate the gray value of each pixel in the grayscale image, and perform average processing on the gray values ​​of all pixels to obtain the average gray value of the current image.

[0123] S303, Set the target desired gray value, calculate the difference between the current image average gray value and the target desired gray value, and obtain the gray value difference;

[0124] S304 analyzes the light source intensity based on the lighting environment, obtains the camera's exposure time, and dynamically adjusts the light source intensity and exposure time based on the grayscale difference to obtain the final display panel image.

[0125] It should be noted that by averaging the gray values ​​of the pixels to obtain the average gray value of the image, and comparing it with the target desired gray value, the light source intensity and exposure time are dynamically adjusted according to the gray value difference to obtain a clear image of the display panel.

[0126] According to an embodiment of the present invention, the image of the display panel is preprocessed to obtain a preprocessed image, specifically including:

[0127] Obtain the final display panel image and mark the pixel coordinates of the four corner points on the display panel image;

[0128] A transformation matrix is ​​constructed based on the four corner calibration points. The ideal coordinates of the corresponding points in the target image are set. The corner pixel coordinates are transformed to the ideal coordinates based on the transformation matrix to obtain the corrected image.

[0129] The corrected image is then normalized to obtain a fixed-size image;

[0130] Based on a fixed-size image, the brightness values ​​of different pixels in the image are analyzed, and the brightness uniformity is analyzed based on the brightness values ​​of all pixels.

[0131] The brightness uniformity is compared with the set uniformity threshold to obtain the uniformity difference. Brightness value compensation is then performed based on the uniformity difference to obtain the preprocessed image.

[0132] It should be noted that the transformation matrix H is constructed using the four corner calibration points:

[0133] ,

[0134] Where (x,y) are the original image coordinates, and (x′,y′) are the corrected coordinates; the image is geometrically reconstructed using bilinear interpolation through a transformation matrix.

[0135] According to an embodiment of the present invention, defect features are extracted from blotchy features and texture features based on a threshold segmentation algorithm to obtain defect information, specifically including:

[0136] Obtain the preprocessed image, and perform grayscale normalization on the preprocessed image to obtain the normalized image;

[0137] Extract normalized image features, set a first segmentation threshold, and segment the normalized image features based on the segmentation threshold to obtain image foreground features and image background features.

[0138] A second segmentation threshold is set, and the foreground features of the image are segmented based on the second segmentation threshold to obtain blob features and texture features;

[0139] A defect threshold is set, and defect feature segmentation is performed on the spot features and texture features based on the defect threshold to obtain defect information.

[0140] It should be noted that the process involves determining whether a Mura defect exists in the image and performing type identification.

[0141] The determination method is as follows:

[0142] (1) Threshold segmentation:

[0143] ,

[0144] This represents the value at pixel location (x, y) in the normalized image, indicating the normalized brightness value.

[0145] The average grayscale value of a local region centered at pixel (x, y) with radius r is called the local brightness reference value.

[0146] : Dynamically set based on local maximum value statistics.

[0147] (2) Defect feature extraction:

[0148] area:

[0149] ,

[0150] Shape factor (circularity):

[0151] P is the perimeter.

[0152] C is the Circularity Factor, which represents the degree of roundness.

[0153] Directionality (used to determine strip-type Mura):

[0154] ,

[0155] The term "directional length" refers to the length of the defect along its principal axis, representing the extent of its extension in that direction, i.e., the "stretching" in the direction in which the defect is most prominent.

[0156] B represents the width of the defect area, referring to the average or minimum width perpendicular to the main direction. It reflects the "narrowness" of strip-type defects.

[0157] The direction variation (θ) or angle spread parameter indicates the range of changes in the direction angle, i.e., the directional consistency of the stripe in space. The smaller the θ, the straighter and more directional the defect. θ may be expressed in radians or standard deviation, etc.

[0158] Defect types include:

[0159] Blocky Mura: Closed outline, large area, and consistent grayscale deviation.

[0160] Mura bands: highly directional, elongated shape;

[0161] Cloud-like Mura: Low-frequency blurred regions, loosely distributed.

[0162] According to an embodiment of the present invention, defect features are classified based on defect information to obtain different defect types, and an inspection report is generated, specifically including:

[0163] Obtain several defect samples including different defect categories, and divide the defect samples into training set and test set according to a set ratio;

[0164] Build an initial model framework, and iteratively train the initial model framework based on the training set to obtain the training results;

[0165] Determine whether the training results have converged;

[0166] If convergence is achieved, the model is tested on the test set, and the hyperparameters of the model are dynamically adjusted according to the test results to obtain the defect classification model. Based on the defect classification model, the defect information is analyzed to obtain different defect types.

[0167] Different defect types are classified and labeled to generate inspection reports;

[0168] If the model does not converge, adjust the number of iterations until it converges.

[0169] It should be noted that by continuously training the model with several defect samples, the output accuracy of the model can be improved. After the model is trained, it is tested to dynamically adjust the hyperparameters of the model, so that the output results of the model are closer to the actual results and the defect classification accuracy can be improved.

[0170] Please refer to Figure 4 , Figure 4 This is a block diagram of a machine vision-based display screen defect detection system according to some embodiments of this application. In a second aspect, embodiments of this application provide a machine vision-based display screen defect detection system, comprising: a memory and a processor. The memory includes a program for a machine vision-based display screen defect detection method. When the program for the machine vision-based display screen defect detection method is executed by the processor, it implements the following steps:

[0171] Set the light source parameters to create the lighting environment for detection, and place the display panel in the lighting environment;

[0172] The image of the display panel is obtained by acquiring the image of the display panel against a high grayscale background using a high dynamic range industrial camera. The display panel image is then preprocessed to obtain a preprocessed image.

[0173] Image features are extracted from the preprocessed image, and blob and texture features are selected from the image based on these features.

[0174] Defect features are extracted from spot and texture features using a threshold segmentation algorithm to obtain defect information.

[0175] Based on the defect information, the defect features are classified to obtain different defect types, and an inspection report is generated.

[0176] Specifically, the defect detection system includes a high-uniformity planar light source 1, a high-dynamic industrial camera 2, an automatic brightness adjustment controller 3, an image acquisition and geometric correction unit 4, a multi-scale spatial frequency enhancement module 5, a LIP logarithmic enhancement module 6, and a Mura defect judgment and classification module 7.

[0177] The highly uniform planar light source 1 provides a uniform, spotless surface illumination environment for the display panel being inspected, avoiding interference from uneven light sources.

[0178] Technical requirements: Illumination uniformity ≥ 95%; color temperature consistency ±50K; supports grayscale / white screen / solid color display backgrounds;

[0179] Principle: It uses an integrating light guide plate + diffusion film structure, combined with a large area LED dimming array.

[0180] The high dynamic range industrial camera 2 captures images of the display panel against a high grayscale background and is required to identify subtle spot differences with a ΔL (brightness difference) as low as 3%.

[0181] Technical requirements: Resolution ≥ 12MP; Supports 12-bit or higher grayscale imaging; Global shutter to avoid image tearing;

[0182] Interface: Supports high-speed data output protocols such as CameraLink, GigE, and USB3.0.

[0183] Automatic brightness adjustment controller 3 automatically adjusts the exposure time and light source intensity based on panel brightness and exposure image information to ensure that the grayscale distribution is located in the optimal recognition area (e.g., 70% mid-gray); image acquisition and geometric correction unit 4 preprocesses the acquired image, including: perspective distortion correction; size normalization; flatness compensation; multi-scale spatial frequency enhancement module 5 extracts low-contrast features such as spots and textures from the image; LIP logarithmic enhancement module 6 further enhances the dark texture of low-contrast images, using the following formula:

[0184] ,

[0185] Symbol explanation: Frequency-enhanced images;

[0186] : Local logarithmic mean;

[0187] : Results after logarithmic image enhancement.

[0188] The Mura defect detection and classification module 7 is used to determine whether there is a Mura defect in the image and to perform type identification.

[0189] According to an embodiment of the present invention, setting light source parameters to form a detection lighting environment and placing the display panel in the lighting environment specifically includes:

[0190] Set the light source parameters, including light source brightness, light source uniformity, and color temperature value;

[0191] Multiple light source parameter combinations are matched based on the ratio of light source brightness, light source uniformity, and color temperature value;

[0192] Multiple lighting environments are formed based on combinations of light source parameters, and the presence of local bright spots in these environments is analyzed.

[0193] If present, adjust the ratio of light source brightness, light source uniformity, and color temperature value;

[0194] If there are no local bright spots, the required lighting environment for testing is obtained. Place the display screen in the center of the lighting environment and adjust the angle and level of the display screen.

[0195] It should be noted that by setting different proportions of light source brightness, light source uniformity, and color temperature, multiple combinations of light source parameters are obtained. By analyzing whether the lighting environment formed by these multiple light source parameter combinations meets the requirements, the proportions of light source brightness, light source uniformity, and color temperature can be precisely adjusted to obtain the optimal lighting environment.

[0196] According to an embodiment of the present invention, an image of the display panel is obtained by acquiring an image of the display panel against a high grayscale background using a high dynamic range industrial camera, specifically including:

[0197] Acquire the original display panel image, and perform grayscale processing on the original display panel image to obtain a grayscale image;

[0198] Calculate the gray value of each pixel in the grayscale image, and average the gray values ​​of all pixels to obtain the average gray value of the current image.

[0199] Set the target desired gray value, calculate the difference between the current image's average gray value and the target desired gray value, and obtain the gray value difference;

[0200] The light source intensity is analyzed based on the lighting environment, the camera exposure time is obtained, and the light source intensity and exposure time are dynamically adjusted based on the grayscale difference to obtain the final display panel image.

[0201] It should be noted that by averaging the gray values ​​of the pixels to obtain the average gray value of the image, and comparing it with the target desired gray value, the light source intensity and exposure time are dynamically adjusted according to the gray value difference to obtain a clear image of the display panel.

[0202] According to an embodiment of the present invention, the image of the display panel is preprocessed to obtain a preprocessed image, specifically including:

[0203] Obtain the final display panel image and mark the pixel coordinates of the four corner points on the display panel image;

[0204] A transformation matrix is ​​constructed based on the four corner calibration points. The ideal coordinates of the corresponding points in the target image are set. The corner pixel coordinates are transformed to the ideal coordinates based on the transformation matrix to obtain the corrected image.

[0205] The corrected image is then normalized to obtain a fixed-size image;

[0206] Based on a fixed-size image, the brightness values ​​of different pixels in the image are analyzed, and the brightness uniformity is analyzed based on the brightness values ​​of all pixels.

[0207] The brightness uniformity is compared with the set uniformity threshold to obtain the uniformity difference. Brightness value compensation is then performed based on the uniformity difference to obtain the preprocessed image.

[0208] It should be noted that the transformation matrix H is constructed using the four corner calibration points:

[0209] ,

[0210] Where (x,y) are the original image coordinates, and (x′,y′) are the corrected coordinates; the image is geometrically reconstructed using bilinear interpolation through a transformation matrix.

[0211] According to an embodiment of the present invention, defect features are extracted from blotchy features and texture features based on a threshold segmentation algorithm to obtain defect information, specifically including:

[0212] Obtain the preprocessed image, and perform grayscale normalization on the preprocessed image to obtain the normalized image;

[0213] Extract normalized image features, set a first segmentation threshold, and segment the normalized image features based on the segmentation threshold to obtain image foreground features and image background features.

[0214] A second segmentation threshold is set, and the foreground features of the image are segmented based on the second segmentation threshold to obtain blob features and texture features;

[0215] A defect threshold is set, and defect feature segmentation is performed on the spot features and texture features based on the defect threshold to obtain defect information.

[0216] It should be noted that the process involves determining whether a Mura defect exists in the image and performing type identification.

[0217] The determination method is as follows:

[0218] (1) Threshold segmentation:

[0219] ,

[0220] : Dynamically set based on local maximum value statistics;

[0221] (2) Defect feature extraction:

[0222] area:

[0223] ,

[0224] Shape factor (circularity):

[0225] P is the perimeter.

[0226] Directionality (used to determine strip-type Mura):

[0227] .

[0228] Defect types include:

[0229] Blocky Mura: Closed outline, large area, and consistent grayscale deviation.

[0230] Mura bands: highly directional, elongated shape;

[0231] Cloud-like Mura: Low-frequency blurred regions, loosely distributed.

[0232] According to an embodiment of the present invention, defect features are classified based on defect information to obtain different defect types, and an inspection report is generated, specifically including:

[0233] Obtain several defect samples including different defect categories, and divide the defect samples into training set and test set according to a set ratio;

[0234] Build an initial model framework, and iteratively train the initial model framework based on the training set to obtain the training results;

[0235] Determine whether the training results have converged;

[0236] If convergence is achieved, the model is tested on the test set, and the hyperparameters of the model are dynamically adjusted according to the test results to obtain the defect classification model. Based on the defect classification model, the defect information is analyzed to obtain different defect types.

[0237] Different defect types are classified and labeled to generate inspection reports;

[0238] If the model does not converge, adjust the number of iterations until it converges.

[0239] It should be noted that by continuously training the model with several defect samples, the output accuracy of the model can be improved. After the model is trained, it is tested to dynamically adjust the hyperparameters of the model, so that the output results of the model are closer to the actual results and the defect classification accuracy can be improved.

[0240] A third aspect of the present invention provides a computer-readable storage medium including a machine vision-based display screen defect detection method program. When the machine vision-based display screen defect detection method program is executed by a processor, it implements the steps of the machine vision-based display screen defect detection method as described in any of the above claims.

[0241] This invention discloses a machine vision-based method, system, and medium for detecting defects in a display screen. By setting light source parameters to create a suitable lighting environment, the display screen panel is placed within this environment. A high dynamic range industrial camera captures images of the display screen panel against a high grayscale background, resulting in a display screen panel image. This image is then preprocessed to obtain a preprocessed image. Image features are extracted from the preprocessed image, and spot and texture features are selected based on these features. A threshold segmentation algorithm is used to extract defect features from the spot and texture features, yielding defect information. Based on this defect information, the defect features are classified into different defect types, and a detection report is generated. By dynamically adjusting the light source parameters of the lighting environment to provide an optimal detection environment, and by segmenting defect features using the threshold segmentation algorithm, the method accurately analyzes and identifies spot defects on the display screen panel, reducing false detection rates and environmental sensitivity, and improving detection accuracy.

[0242] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0243] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0244] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0245] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0246] Alternatively, if the integrated units of the present invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of the present invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This 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 methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

Claims

1. A method for detecting defects in a display screen based on machine vision, characterized in that, The method comprises the following steps: setting light source parameters, including light source brightness, light source uniformity and color temperature value; matching the proportion value of light source brightness, light source uniformity and color temperature value to form multiple light source parameter combinations; forming multiple lighting environments based on the light source parameter combinations, and analyzing whether there are local bright spots in the lighting environments; if there are, adjusting the proportion value of light source brightness, light source uniformity and color temperature value; if there are no local bright spots, obtaining the required lighting environment for detection, setting the display screen at the center position of the lighting environment, and adjusting the angle and levelness of the display screen; based on a high dynamic industrial camera, collecting an image of the display screen panel under a high gray scale background to obtain a display screen panel image, and preprocessing the display screen panel image to obtain a preprocessed image; based on the preprocessed image, extracting image features, and based on the image features, screening out spot features and texture features in the image; based on a threshold segmentation algorithm, extracting defect features from the spot features and texture features to obtain defect information; based on the defect information, classifying the defect features to obtain different defect types, and generating a detection report; wherein, based on the threshold segmentation algorithm, the defect features are extracted from the spot features and texture features to obtain the defect information, specifically including: obtaining the preprocessed image, performing gray scale normalization processing on the preprocessed image to obtain a normalized image; extracting normalized image features, setting a first segmentation threshold, and based on the segmentation threshold, segmenting the normalized image features to obtain image foreground features and image background features; setting a second segmentation threshold, and based on the second segmentation threshold, segmenting the image foreground features to obtain spot features and texture features; setting a defect threshold, and based on the defect threshold, segmenting the spot features and texture features to obtain defect information.

2. The machine vision-based display screen defect detection method of claim 1, wherein, Based on a high dynamic industrial camera, an image of the display screen panel under a high gray scale background is collected to obtain a display screen panel image, specifically including: obtaining an original display screen panel image, performing gray value processing on the original display screen panel image to obtain a gray scale image; calculating the gray scale value of each pixel point in the gray scale image, and performing mean value processing on the gray scale values of all pixel points to obtain the current image average gray scale value; setting a target expected gray scale value, and calculating the difference between the current image average gray scale value and the target expected gray scale value to obtain a gray scale difference value; based on the light source brightness of the lighting environment, analyzing the light source intensity, obtaining the exposure time of the camera, and dynamically adjusting the light source intensity and the exposure time based on the gray scale difference value to obtain the final display screen panel image.

3. The machine vision-based display screen defect detection method of claim 2, wherein, The display screen panel image is preprocessed to obtain a preprocessed image, specifically including: obtaining the final display screen panel image, and marking four corner pixel coordinates on the display screen panel image; based on the four corner calibration points, constructing a transformation matrix, setting ideal coordinates of corresponding points in the target image, and converting the corner pixel coordinates to the ideal coordinates based on the transformation matrix to obtain a corrected image; performing size normalization processing on the corrected image to obtain a fixed size image; based on the fixed size image, analyzing the brightness values of different pixel points of the image, and analyzing the brightness uniformity according to the brightness values of all pixel points; The uniformity difference value is obtained by comparing the brightness uniformity with the set uniformity threshold value, the brightness value is compensated according to the uniformity difference value, and the preprocessed image is obtained.

4. The machine vision-based display screen defect detection method of claim 3, wherein, The defect features are classified based on the defect information to obtain different defect types, and a detection report is generated, specifically including: A plurality of defect samples including different defect categories are obtained, and the defect samples are divided into a training set and a test set according to a set proportion; An initial model framework is constructed, and the initial model framework is iteratively trained based on the training set to obtain a training result; It is judged whether the training result converges or not; If the training result converges, the model is tested based on the test set, the hyperparameters of the model are dynamically adjusted according to the test result, a defect classification model is obtained, and the defect information is analyzed based on the defect classification model to obtain different defect types; Different defect types are classified and labeled to generate a detection report; If the training result does not converge, the number of iterations is adjusted until the model converges.

5. A machine vision based display panel defect detection system, characterized in that, The system includes a memory and a processor, the memory includes a machine vision-based display screen defect detection method program, and the machine vision-based display screen defect detection method program is executed by the processor to implement the following steps: Set the light source parameters, including light source brightness, light source uniformity and color temperature value; Match the proportion value of the light source brightness, light source uniformity and color temperature value to form a plurality of light source parameter combinations; Form a plurality of lighting environments based on the light source parameter combinations, and analyze whether there are local bright spots in the lighting environment; If there are, adjust the proportion value of the light source brightness, light source uniformity and color temperature value; If there are no local bright spots, obtain the required lighting environment for detection, set the display screen at the center position of the lighting environment, and adjust the angle and level of the display screen; Collect the image of the display panel under the high gray scale background based on the high dynamic industrial camera to obtain the display panel image, and preprocess the display panel image to obtain the preprocessed image; Extract image features based on the preprocessed image, and screen out spot features and texture features in the image based on the image features; Extract defect features based on the threshold segmentation algorithm to obtain defect information; Classify the defect features based on the defect information to obtain different defect types, and generate a detection report; Wherein, the defect features are extracted based on the threshold segmentation algorithm to obtain the defect information, specifically including: Obtain the preprocessed image, and perform gray scale normalization processing on the preprocessed image to obtain a normalized image; Extract the normalized image features, set a first segmentation threshold, and segment the normalized image features based on the segmentation threshold to obtain image foreground features and image background features; Set a second segmentation threshold, and segment the image foreground features based on the second segmentation threshold to obtain spot features and texture features; Set a defect threshold, and segment the spot features and texture features based on the defect threshold to obtain defect information.

6. The machine vision-based display screen defect detection system of claim 5, wherein, Collect the image of the display panel under the high gray scale background based on the high dynamic industrial camera to obtain the display panel image, specifically including: Obtain the original display panel image, and perform gray value processing on the original display panel image to obtain a gray scale image; The gray value of each pixel point in the gray image is calculated, and the gray values of all pixel points are processed by mean value to obtain a current image average gray value; A target expected gray value is set, and a difference value is calculated by the current image average gray value and the target expected gray value to obtain a gray difference value; The light source intensity is analyzed based on the light source brightness of the lighting environment, the exposure time of the camera is acquired, the light source intensity and the exposure time are dynamically adjusted based on the gray difference value, and a final display screen panel image is obtained.

7. A computer-readable storage medium, characterized in that, The computer readable storage medium comprises a display screen defect detection method based on machine vision program, and the display screen defect detection method based on machine vision program is executed by the processor to realize the steps of the display screen defect detection method based on machine vision in any one of claims 1 to 4.

Images