A display panel detection method

Abstract

This invention discloses a display panel detection method, belonging to the field of target detection technology. The method includes acquiring multiple original display panel images from different viewpoints; preprocessing the acquired original display panel images to extract features of suspected defect areas in each preprocessed image; aligning the spatial coordinates of suspected defect areas in images from different viewpoints to determine whether defects captured from different viewpoints are the same physical defect, and establishing a defect feature vector containing feature parameters of the same defect under different viewpoints; inputting the defect feature vector into a preset judgment model, and performing a comprehensive judgment based on the judgment model to determine the final category of the defect. This invention employs the above-mentioned display panel detection method, which improves the accuracy of defect detection and reduces the false positive rate through multi-view synchronous imaging and multi-feature fusion judgment.

Description

Technical Field

[0001] This invention relates to the field of target detection technology, and in particular to a method for detecting display panels. Background Technology

[0002] As a core display component for human-computer interaction, the display panel's display quality directly determines the user experience and market competitiveness of end products. To ensure factory quality, display panels must undergo rigorous defect detection and grading, which are indispensable key processes in the production process.

[0003] In existing technologies, display panel defect detection mainly relies on two methods: manual inspection and automated inspection. While manual inspection can perceive defects by switching between multiple viewing angles, it suffers from low efficiency, strong subjectivity, and susceptibility to fatigue, making it difficult to meet the demands of high-speed automated production lines. Automated inspection, on the other hand, uses a single-view or fixed frontal viewing angle imaging scheme, acquiring frontal images for defect identification and judgment. Although this automates and standardizes the inspection process, display panel defects vary in shape, and different defects appear differently from different viewing angles. Some defects are not obvious from the frontal view but are obvious from side or other views, while others are obvious from the frontal view but not from other angles. Making a final judgment based solely on the defect shape from the frontal view will introduce judgment errors. Summary of the Invention

[0004] The purpose of this invention is to provide a display panel detection method to solve the problems in the background art.

[0005] To achieve the above objectives, the present invention provides a display panel detection method, comprising the following steps: S1. Control multiple image acquisition units arranged at different angles to synchronously image the display panel under inspection at the same time and acquire multiple original images from different angles. S2. Preprocess the acquired original image of the display panel and extract the features of the suspected defect area in each image after preprocessing. S3. Align the spatial coordinates of suspected defect areas in images from different perspectives, determine whether the defects captured from different perspectives are the same physical defects, and establish a defect feature vector containing the feature parameters of the same defect from different perspectives. S4. Input the defect feature vector into the preset judgment model, make a comprehensive judgment based on the judgment model, and determine the final category of the defect.

[0006] Preferably, in step S1, multiple image acquisition units are arranged above and around the display panel, respectively. Each image acquisition unit includes a lens, a CCD camera, and a fill light assembly. The lens is placed in front of the CCD camera lens, and the fill light assembly adopts a uniform fill light mode for fill light.

[0007] Preferably, the preprocessing includes image noise reduction, contrast enhancement, and geometric correction.

[0008] Preferably, the features extracted from the suspected defect areas in each preprocessed image include: An adaptive threshold segmentation algorithm is used to segment the preprocessed image, and regions with gray values ​​exceeding the normal range are selected as suspected defect regions. The suspected defect regions are then further filtered. In the adaptive threshold segmentation algorithm, the threshold is dynamically adjusted according to the gray value distribution of the image. The location coordinates, morphological features and optical features of each suspected defect area after screening are extracted. The morphological features include size, shape factor and orientation, and the optical features include gray value, edge contour, texture features and contrast. After extraction, a feature profile is created for each suspected defect region in each viewpoint image, marking the viewpoint information, defect number, and corresponding feature parameters.

[0009] Preferably, step S3 specifically includes: S31. Based on the frontal view image, establish a three-dimensional spatial coordinate system for the display panel. Based on the image registration algorithm, calculate the transformation matrix of different view images relative to the base image. According to the transformation matrix, transform the coordinates of the suspected defect area under different view angles, and perform spatial alignment according to the transformed coordinates. S32. For suspected defect areas from all perspectives, determine whether they are the same physical defect based on spatial coordinate distance and feature similarity. S33. For each physical defect after judgment, integrate its features from all perspectives to establish a defect feature vector.

[0010] Preferably, step S32 specifically involves: setting a spatial distance threshold and a feature similarity threshold; if two suspected defect areas from different perspectives have a spatial coordinate distance less than the threshold and a feature similarity greater than the threshold, they are determined to be the same physical defect; if suspected defect areas from multiple perspectives meet the conditions, they are merged into the same physical defect, and suspected defects that are repeatedly determined are eliminated.

[0011] Preferably, step S4 specifically involves: inputting each defect feature vector into a preset judgment model; the judgment model performs comprehensive reasoning through multi-dimensional analysis of the feature vectors; if the judgment confidence is greater than a preset threshold, it outputs a clear defect category, defect level, and judgment confidence; if the judgment confidence does not exceed the preset threshold, it is marked as a defect to be inspected and confirmed through manual review.

[0012] Preferably, in step S4, the model is determined to be a trained SVM support vector machine model.

[0013] Therefore, the present invention adopts the above-mentioned display panel detection method, which improves the accuracy of defect detection and reduces the false judgment rate by using multi-view synchronous imaging and multi-feature fusion judgment.

[0014] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0015] Figure 1 This is a flowchart of a display panel detection method according to an embodiment of the present invention; Figure 2 This is a diagram showing the placement of the image acquisition unit in a display panel detection method according to an embodiment of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention 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 the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0017] Example like Figures 1-2 As shown, the present invention provides a display panel detection method, including the following steps: S1. Control multiple image acquisition units arranged at different viewing angles to simultaneously image the display panel under inspection, acquiring multiple original images from different perspectives. Five image acquisition units are configured. Taking a flat-lying display panel as an example, the image acquisition units are arranged directly above and around the display panel, and each unit is required to cover the entire display area. Each image acquisition unit includes a lens, a CCD camera, and a supplementary lighting assembly. The lens is placed in front of the CCD camera lens, and the supplementary lighting assembly uses a uniform illumination mode for supplementary lighting.

[0018] S2. Preprocess the acquired original images of the display panel, extracting features of suspected defect areas from each preprocessed image. Preprocessing includes image denoising, contrast enhancement, and geometric correction. Image denoising uses a Gaussian filtering algorithm to filter random noise in the image, preserving detailed information in defect areas and preventing noise from being misidentified as suspected defects. Contrast enhancement uses a histogram equalization algorithm to adjust the grayscale range of the image, increasing the grayscale difference between defect areas and normal display areas, making subtle defects clearly visible. Geometric correction ensures accurate subsequent coordinate alignment.

[0019] Features of suspected defect areas extracted from each preprocessed image include: An adaptive thresholding segmentation algorithm is used to segment the preprocessed image, filtering out regions with grayscale values ​​exceeding the normal range as suspected defect areas. In this algorithm, the threshold is dynamically adjusted based on the image's grayscale distribution. Contour extraction is performed on each suspected defect area, eliminating tiny regions smaller than a preset threshold and retaining only the valid suspected defect areas.

[0020] The location coordinates, morphological features, and optical features of each suspected defect area after screening are extracted. The morphological features include features such as size, shape factor, and orientation. The shape factor is used to distinguish point defects, line defects, and surface defects. The optical features include features such as gray value, edge contour, texture features, and contrast.

[0021] After extraction, a feature profile is created for each suspected defect region in each viewpoint image, marking the viewpoint information, defect number, and corresponding feature parameters.

[0022] S3. Align the spatial coordinates of suspected defect areas in images from different viewpoints, determine whether the defects captured from different viewpoints are the same physical defect, and establish a defect feature vector containing the feature parameters of the same defect from different viewpoints. Specifically, this includes: S31. Based on the frontal view image, establish a three-dimensional spatial coordinate system for the display panel. Using an image registration algorithm, extract feature matching points between each view image and the reference image. Calculate the transformation matrix of each view image relative to the base image. Based on the transformation matrix, convert the two-dimensional coordinates of suspected defect areas under each view into three-dimensional spatial coordinates. Perform spatial alignment based on the transformed coordinates. In the three-dimensional spatial coordinate system, the X-axis represents the horizontal direction of the panel, the Y-axis represents the vertical direction of the panel, and the Z-axis represents the direction perpendicular to the panel surface.

[0023] S32. For suspected defect areas from all viewpoints, determine whether they belong to the same physical defect based on spatial coordinate distance and feature similarity. Specifically, this includes: Set a spatial distance threshold and a feature similarity threshold. If two suspected defect areas from different perspectives have a spatial coordinate distance less than the threshold and a feature similarity greater than the threshold, they are determined to be the same physical defect. If suspected defect areas from multiple perspectives meet the conditions, they are merged into the same physical defect, and suspected defects that are repeatedly judged are removed.

[0024] S33. For each identified physical defect, integrate its features from all perspectives to establish a defect feature vector. Simultaneously, assign a unique defect label to each defect, linking it to the original images and feature parameters from all perspectives for easy subsequent tracing.

[0025] S4. Input the defect feature vector into the preset judgment model, and make a comprehensive judgment based on the judgment model to determine the final category of the defect. Specifically, this includes: Each defect feature vector is input into a preset judgment model. The judgment model performs comprehensive reasoning through multi-dimensional analysis of the feature vectors. If the judgment confidence is greater than the preset threshold, it outputs a clear defect category, defect level, and judgment confidence (minor, moderate, severe). If the judgment confidence does not exceed the preset threshold, it is marked as a defect to be inspected and confirmed by manual review.

[0026] The judgment model is a pre-trained SVM support vector machine model, which is trained, validated, and optimized using a large amount of labeled data covering various common defects and defect features from different perspectives to ensure the accuracy of the judgment model.

[0027] To verify the effectiveness of the present invention, a comparative experiment was conducted. As shown in Table 1, the results of the comparison between the method of the present invention and the method of defect judgment based solely on frontal view images are as follows: The table shows that the accuracy of the present invention is significantly improved and the false negative rate is significantly reduced.

[0028] Table 1 Comparison Results

[0029] The present invention also provides a display panel detection system for performing the above-described display panel detection method.

[0030] The present invention also provides a storage medium on which a computer program is stored, which, when executed, can perform the steps provided in the above embodiments. The storage medium may include various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0031] The present invention also provides an electronic device, including a memory and a processor. The memory stores a computer program, and when the processor calls the computer program in the memory, it can implement the steps provided in the above embodiments.

[0032] Therefore, the present invention adopts the above-mentioned display panel detection method, which improves the accuracy of defect detection and reduces the false judgment rate by using multi-view synchronous imaging and multi-feature fusion judgment.

[0033] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for detecting a display panel, characterized in that, Including the following steps: S1. Control multiple image acquisition units arranged at different angles to synchronously image the display panel under inspection at the same time and acquire multiple original images from different angles. S2. Preprocess the acquired original image of the display panel and extract the features of the suspected defect area in each image after preprocessing. S3. Align the spatial coordinates of suspected defect areas in images from different perspectives, determine whether the defects captured from different perspectives are the same physical defects, and establish a defect feature vector containing the feature parameters of the same defect from different perspectives. S4. Input the defect feature vector into the preset judgment model, make a comprehensive judgment based on the judgment model, and determine the final category of the defect.

2. The display panel detection method according to claim 1, characterized in that: In step S1, multiple image acquisition units are arranged above and around the display panel. Each image acquisition unit includes a lens, a CCD camera, and a fill light assembly. The lens is placed in front of the CCD camera lens, and the fill light assembly uses a uniform fill light mode for fill light.

3. The display panel detection method according to claim 1, characterized in that: Preprocessing includes image noise reduction, contrast enhancement, and geometric correction.

4. The display panel detection method according to claim 1, characterized in that, Features of suspected defect areas extracted from each preprocessed image include: An adaptive threshold segmentation algorithm is used to segment the preprocessed image, and regions with gray values ​​exceeding the normal range are selected as suspected defect regions. The suspected defect regions are then further filtered. In the adaptive threshold segmentation algorithm, the threshold is dynamically adjusted according to the gray value distribution of the image. The location coordinates, morphological features and optical features of each suspected defect area after screening are extracted. The morphological features include size, shape factor and orientation, and the optical features include gray value, edge contour, texture features and contrast. After extraction, a feature profile is created for each suspected defect region in each viewpoint image, marking the viewpoint information, defect number, and corresponding feature parameters.

5. The display panel detection method according to claim 1, characterized in that, Step S3 specifically includes: S31. Based on the frontal view image, establish a three-dimensional spatial coordinate system for the display panel. Based on the image registration algorithm, calculate the transformation matrix of different view images relative to the base image. According to the transformation matrix, transform the coordinates of the suspected defect area under different view angles, and perform spatial alignment according to the transformed coordinates. S32. For suspected defect areas from all perspectives, determine whether they are the same physical defect based on spatial coordinate distance and feature similarity. S33. For each physical defect after judgment, integrate its features from all perspectives to establish a defect feature vector.

6. The display panel detection method according to claim 5, characterized in that, Step S32 specifically involves setting a spatial distance threshold and a feature similarity threshold. If two suspected defect areas from different perspectives have a spatial coordinate distance less than the threshold and a feature similarity greater than the threshold, they are determined to be the same physical defect. If suspected defect areas from multiple perspectives meet the conditions, they are merged into the same physical defect, and suspected defects that are repeatedly determined are eliminated.

7. The display panel detection method according to claim 1, characterized in that, Step S4 is as follows: Each defect feature vector is input into a preset judgment model. The judgment model performs comprehensive reasoning through multi-dimensional analysis of the feature vectors. If the judgment confidence is greater than a preset threshold, it outputs a clear defect category, defect level, and judgment confidence. If the judgment confidence does not exceed the preset threshold, it is marked as a defect to be inspected and confirmed through manual review.

8. The display panel detection method according to claim 1, characterized in that: In step S4, the model is determined to be a trained SVM support vector machine model.

Images