Display panel defect detection method, device, equipment, medium and product based on attention potential

Abstract

The present invention discloses a method, apparatus, device, medium, and product for display panel defect detection based on attention potential energy. The method comprises: obtaining a multi-scale channel image set associated with the panel to be inspected; assigning a mass value to each image pixel in the multi-scale channel image set to obtain a point mass volume set for the multi-scale channel; constructing an attention potential energy function and obtaining an attention potential energy vector set for the multi-scale channel based on the point mass volume set for the multi-scale channel; and obtaining a defect detection result for the panel to be inspected based on the attention potential energy value and attention potential energy range associated with the attention potential energy vector set for the multi-scale channel. The present invention ensures that the defect detection result is highly aligned with the human eye, thereby reducing distortion of the detection result.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial machine vision defect detection, and in particular to a display panel defect detection method, device, equipment, medium and product based on attention potential energy. Background Art

[0002] With the development of the general trend of "machines replacing humans", the application of machine vision systems in the industrial chain is becoming more and more extensive. In the field of defect detection of luminous panels (including but not limited to LCD, OLED, Micro-OLED, backlight panels and other display devices and their corresponding process products), due to various reasons such as rising manual inspection costs, long training cycles for inspection operators, and human eye damage (occupational diseases) caused by inspection operations, various inspection systems based on machine vision technology are also being applied and have achieved certain results. Stable and reliable luminous surface machine vision systems with inspection results that maintain a high degree of consistency with manual inspection are of great significance to the development of the industry. Among them, the detection algorithm has a core value in the entire inspection system.

[0003] Currently, relevant detection algorithms are mainly divided into three categories: methods based on filtering and image segmentation; methods based on background fitting or modeling; and sample learning methods based on various artificial neural networks. Methods based on filtering and image segmentation mainly achieve the purpose of defect detection through various types of filtering (such as Gabor filtering, "frequency domain contrast sensitivity function filter template", etc.) and segmentation algorithms ("CV model segmentation", "adaptive threshold segmentation", "level set", etc.) and assist other image optimization processing methods. They are classic image algorithms. Methods based on background fitting or modeling mainly obtain the background through various background fitting or modeling methods (such as "using the PCA algorithm to learn a large number of defect-free samples, automatically extracting the difference features between the background and the target, and reconstructing the background image", "double N-order polynomial surface fitting model", "B-spline fitting", etc.), and then achieve the purpose of defect detection through image difference and assisting other image optimization processing methods. Sample learning methods based on various artificial neural networks mainly use various artificial neural networks (such as "PA FPN feature enhancement network", "MobileNetV3 algorithm model", "manually simulate various types of mura defects using graphics processing methods to form a sample library; train models through deep neural networks") and assist with some detailed processing to train defect samples and then make predictions to achieve the purpose of defect detection. This belongs to the traditional deep learning type.

[0004] While these methods can achieve certain results in various applications, they are inherently difficult to detect in illuminated panels. For example, MURA detection, when overlaid with moiré and pixel texture, has extremely low contrast, lacks clear boundaries, and exhibits diverse morphological variations. This results in poor performance from these image processing methods, making effective segmentation difficult. Deep learning-based methods often experience training failures or overfitting, while auxiliary enhancement methods introduce a large amount of noise blocks of the same grayscale, resulting in poor prediction. Furthermore, even if some MURA can be segmented or predicted, the detection results often lack alignment with human visual detection, leading to distortion. Summary of the Invention

[0005] The present invention provides a display panel defect detection method, device, equipment, medium and product based on attention potential energy to solve the technical problem of defect detection distortion caused by poor alignment between detection effect and human eye detection in the prior art.

[0006] According to one aspect of the present invention, a method for detecting display panel defects based on attention potential is provided, comprising:

[0007] Obtain an image set of multi-scale channels associated with the panel to be detected;

[0008] Assigning a mass value to each image pixel in the image set of the multi-scale channel to obtain a point mass volume set of the multi-scale channel;

[0009] Constructing an attention potential energy function, and obtaining an attention potential energy vector set of the multi-scale channel according to the point mass volume set of the multi-scale channel;

[0010] A defect detection result of the panel to be inspected is obtained based on the attention potential energy value and the attention potential energy range associated with the attention potential energy vector set of the multi-scale channel.

[0011] According to another aspect of the present invention, a display panel defect detection device based on attention potential is provided, comprising:

[0012] A first acquisition module is used to acquire an image set of multi-scale channels associated with the panel to be detected;

[0013] A first conversion module is configured to assign a quality value to each image pixel in the image set of the multi-scale channel to obtain a point mass volume set of the multi-scale channel;

[0014] A second conversion module is used to construct an attention potential energy function and obtain an attention potential energy vector set of the multi-scale channel according to the point mass volume set of the multi-scale channel;

[0015] A detection module is used to obtain a defect detection result of the panel to be inspected based on the attention potential energy value and the attention potential energy range associated with the attention potential energy vector set of the multi-scale channel.

[0016] According to another aspect of the present invention, an electronic device is provided, comprising:

[0017] at least one processor; and

[0018] a memory communicatively connected to the at least one processor; wherein,

[0019] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the display panel defect detection method based on attention potential energy described in any embodiment of the present invention.

[0020] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the display panel defect detection method based on attention potential energy described in any embodiment of the present invention when executed.

[0021] According to another aspect of the present invention, a computer program product is provided, which includes a computer program. When the computer program is executed by a processor, it implements the display panel defect detection method based on attention potential energy described in any embodiment of the present invention.

[0022] The technical solution of the embodiment of the present invention assigns a mass value to each image pixel contained in each image in the image set of the multi-scale channel associated with the panel to be inspected, thereby obtaining a point mass volume set of the multi-scale channel, and simulating the characteristics of the human eye by constructing an attention potential energy function, and obtaining an attention potential energy vector set of the multi-scale channel based on the point mass volume set of the multi-scale channel, and obtaining the defect detection result of the panel to be inspected based on the attention potential energy value associated with the attention potential energy vector set of the multi-scale channel, so that the defect detection result has a highly aligned relationship with the human eye, thereby reducing the distortion of the detection result.

[0023] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0025] Figure 1 is a flow chart of a display panel defect detection method based on attention potential energy provided by an embodiment of the present invention;

[0026] Figure 2 is a flow chart of another display panel defect detection method based on attention potential energy provided by an embodiment of the present invention;

[0027] Figure 3 This is a flow chart of configuring a state observation table provided by an embodiment of the present invention;

[0028] Figure 4 This is a flowchart for implementing display panel defect detection based on attention potential energy provided by an embodiment of the present invention;

[0029] Figure 5 This is a schematic diagram showing an original image with a strong defect and its attention potential energy vector diagram provided by an embodiment of the present invention;

[0030] Figure 6 This is a schematic diagram showing an original image of a dirt defect and its attention potential energy vector diagram provided by an embodiment of the present invention;

[0031] Figure 7 1 is a schematic diagram showing an original image of a vertical MURA defect and its attention potential energy vector diagram provided by an embodiment of the present invention;

[0032] Figure 8 1 is a schematic diagram showing an original image of a black GAP defect and its attention potential energy vector diagram provided by an embodiment of the present invention;

[0033] Figure 9 1 is a schematic structural diagram of a display panel defect detection device based on attention potential energy provided by an embodiment of the present invention;

[0034] Figure 10 This is a structural block diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0035] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0036] It should be noted that the terms "first", "second", etc. in the description and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0037] In one embodiment, Figure 1 This is a flowchart of a display panel defect detection method based on attention potential energy provided by an embodiment of the present invention. This embodiment is applicable to the situation where defects of a display panel are detected based on attention potential energy. The method can be executed by a display panel defect detection device based on attention potential energy. The display panel defect detection device based on attention potential energy can be implemented in the form of hardware and / or software. The display panel defect detection device based on attention potential energy can be configured in an electronic device.

[0038] It should be noted that the direct function of the display panels being inspected is to display content for human viewing. The quality of their images (especially MURA determination) is essentially based on the human eye's perception. Therefore, the ultimate and sole criterion for evaluating the performance of panel inspection machine vision systems must be alignment with the human eye. It is important to note that "alignment with the human eye" specifically refers to the consistency of inspection results by professionally trained visual inspectors during the display panel manufacturing process. These visual inspection results are inherently highly standardized and consistent, and serve as the final inspection standard for product shipments.

[0039] like Figure 1 As shown, the method includes:

[0040] S110 , obtaining an image set of multi-scale channels associated with the panel to be detected.

[0041] In one example, the panel to be inspected refers to a display light-emitting panel that requires defect inspection; the display panel refers to the core component of the display screen in each terminal device, which is used for the image display function; for example, the terminal device may include but not display one of the following: a computer, a tablet computer, an iPad, a mobile phone, and a television. In one example, a multi-scale channel refers to multiple different physical scales; an image set of a multi-scale channel refers to a collection of images of multiple different physical scales. In one example, the display panel may include, but is not limited to, one of the following: a liquid crystal display (LCD), an organic light-emitting diode (OLED), a micro-OLED (a miniaturized version of OLED), etc.

[0042] In actual operation, the physical scales of specific defects on the panel to be inspected vary considerably, and these differences are relatively large. For example, specific defectivity may include, but is not limited to, one of the following: point, line, spot, MURA, and unevenness; and its physical scale may include different levels, such as micrometers, millimeters, and centimeters. In an embodiment, a scale separation operation can be performed on the target image associated with the panel to be inspected using methods such as a "Laplacian pyramid" to obtain an image set of multi-scale channels, which helps improve the timeliness of subsequent processing steps. For example, assuming that the multi-scale channels include three physical scales: micrometers, millimeters, and centimeters, the target image can be scale-separated using a "Laplacian pyramid" approach to obtain an image set of three physical scales, i.e., the image set includes images of the three physical scales. Specifically, microscopic and subtle defects can be detected based on micrometer-level images; slightly larger defects can be detected based on millimeter-level images; and even larger defects can be detected based on centimeter-level images.

[0043] It should be noted that the image size corresponding to each scale channel is different, but its coverage is the same as the original image.

[0044] S120 , assigning a quality value to each image pixel in the image set of the multi-scale channel to obtain a point mass volume set of the multi-scale channel.

[0045] The point mass volume set of a multi-scale channel refers to a collection of point mass volumes at multiple different physical scales. A point mass volume set refers to the collection of point mass volumes for a scale channel, i.e., the collection of all point mass volumes corresponding to an image of a scale channel. In one example, each image pixel in each image in the multi-scale channel image set can be constructed as a point mass volume containing mass, and a mass value can be assigned to each point mass volume to obtain the point mass volume set of the multi-scale channel.

[0046] S130. Construct an attention potential energy function, and obtain an attention potential energy vector set of the multi-scale channel based on the point mass volume set of the multi-scale channel.

[0047] In one example, the attention potential energy function is constructed based on a pre-constructed potential energy function and an attention kernel function; the attention potential energy function can convert the point mass volume set of the above-mentioned multi-scale channel into a corresponding attention potential energy vector set. The attention potential energy vector set can use the original pixels of the target image as the basic material, and through the potential energy function and the attention kernel function and the integration of the subjective preference factors of the human eye, it has a strong alignment with the detection results of the person.

[0048] S140 , obtaining a defect detection result of the panel to be inspected based on the attention potential energy value and the attention potential energy range associated with the attention potential energy vector set of the multi-scale channel.

[0049] In one example, the defect detection result is used to characterize whether there are defects on the panel to be inspected, as well as a description of related information such as the defect type, defect degree, and defect location; whether there are defects is the basic judgment result of the panel to be inspected. For example, if there are no defects on the panel to be inspected, it is characterized that the panel to be inspected is qualified; if there are defects on the panel to be inspected, it is characterized that the panel to be inspected is unqualified. Of course, if the qualification requirements of the panel to be inspected are not high, the defect detection result of the panel to be inspected can be further judged based on information such as the defect type, defect degree, and defect location when there are defects on the panel to be inspected. For example, if the defect type is a bad pixel, and the defect location is located all around, and the defect degree is that there is one bad pixel, then if the defect detection requirements are not high or the defect parameters are within the detection specification range, the defect detection result can be judged to be qualified.

[0050] In one example, the attention potential energy value, also known as the observation intensity value, can be used to evaluate the final defect detection result of the panel to be inspected. Adaptive thresholding can be performed on the attention potential energy vector set for each scale channel to obtain the corresponding attention potential energy value and attention potential energy range. Based on the attention potential energy value and attention potential energy range, the defect detection result of the panel to be inspected can be determined.

[0051] The technical solution of this embodiment assigns a mass value to each image pixel contained in each image in the image set of the multi-scale channel associated with the panel to be inspected, thereby obtaining a point mass volume set of the multi-scale channel, and simulating the characteristics of the human eye by constructing an attention potential energy function. The attention potential energy vector set of the multi-scale channel is converted based on the point mass volume set of the multi-scale channel, and the defect detection result of the panel to be inspected is obtained based on the attention potential energy value associated with the attention potential energy vector set of the multi-scale channel, so that the defect detection result has a highly aligned relationship with the human eye, thereby reducing the distortion of the detection result.

[0052] In one embodiment, Figure 2 This is a flowchart of another display panel defect detection method based on attention potential energy provided by an embodiment of the present invention. This embodiment, based on the above embodiment, further explains the process of acquiring the image set of the multi-scale channel, the process of determining the point mass volume set of the multi-scale channel, the process of determining the attention potential energy vector set of the multi-scale channel, and the process of determining the defect detection result. Figure 2 As shown, the method includes:

[0053] S210: Acquire an original image associated with the panel to be inspected.

[0054] The original image refers to the initial image of the panel to be inspected, captured using an image acquisition device, before any defect detection algorithms or modifications are performed. Generally speaking, the original image can truly reflect the surface appearance and internal structure of the panel to be inspected (within the image's reproducible range). For example, the image acquisition device can be an industrial camera, and can be a black and white camera, meaning the original image can be a black and white image.

[0055] S220: Perform image preprocessing on the original image to obtain a related target image.

[0056] In one example, image preprocessing refers to some necessary denoising, filtering, and other operations performed on the original image; the target image refers to the image obtained after denoising, filtering, and other operations are performed on the original image. Image preprocessing of the original image can remove or reduce the pixel texture of the panel to be inspected (excluding the backlight panel) and the moiré pattern generated by its imaging process. Although the above texture has obvious periodicity, considering that the texture changes caused by different products and different camera resolutions are very obvious, the filtering effect of frequency domain processing based on Fourier transform is not good, with large frequency residues, complex parameter adjustment, and high computational efficiency. Conventional filtering methods such as Gaussian filtering have poor effects on removing the above texture. Therefore, in this solution, a custom convolution operator is used to perform spatial filtering on the original image, which has the characteristics of high computational efficiency and good filtering effect.

[0057] S230 , performing a scale separation operation on the target image to obtain an image set of multi-scale channels.

[0058] The physical scales of the specific defects of the panel to be inspected, such as points, lines, spots, MURA, and unevenness, vary greatly, ranging from microns to centimeters. Therefore, performing scale separation operations on the target image and obtaining an image set of multi-scale channels is beneficial to improving the timeliness of subsequent processing steps. Generally, methods such as the "Laplacian pyramid" are used to obtain images of corresponding scales through graphics sampling.

[0059] S240 , obtaining each image pixel of each image in the image set of the multi-scale channel.

[0060] In one example, the image pixels contained in each image may be acquired in sequence by traversing pixel by pixel.

[0061] S250 , multiply each image pixel by the associated quality weight coefficient to obtain a point mass volume set of a multi-scale channel.

[0062] In one example, each image pixel contained in each image in the above-mentioned multi-scale channel image set can be regarded as a point mass body containing mass and a mass value can be assigned. For example, the point mass body of the pixel coordinate (i, j) is: g(i, j) = h(i, j) * I(i, j); wherein h(i, j) is a quality weight coefficient that changes according to the position coordinate (i, j), and can be matched and assigned according to the characteristics of different product models and the subjective requirements of quality evaluators. For example, the value range of h(i, j) can be 0-10. Generally, the value of h(i, j) is 1. This realizes the conversion of the multi-scale channel image set into the multi-scale channel point mass body set.

[0063] S260. Construct the potential energy function of a point mass body.

[0064] In one embodiment, the potential energy function includes the following: Among them, G(i,j) represents the point mass body at the pixel coordinate (i,j) within the field radius range k represents the deviation length between i and x-axis; l represents the deviation length between j and y-axis; C represents the perturbation coefficient (used for special case processing, such as the singular point and singular region processing described below, and generally takes a value of 0 to 10).

[0065] S270. Apply the potential energy function to the point mass set of the multi-scale channel to obtain the potential energy vector set of the multi-scale channel.

[0066] Clearly, the potential energy of a point mass is inversely proportional to the square of the domain radius and directly proportional to the mass within the domain. Integration (discrete accumulation) effectively characterizes the "condensed" state of characteristic quantities within a local area. This is highly targeted for characterizing MURA-like defects in light-emitting panels. MURA images alone exhibit very low grayscale differences (1–2 grayscales), lack boundary contours, and exhibit randomly varying morphologies. Processing them pixel by pixel in the original image is nearly impossible, or the resulting image will differ significantly from the human eye's perception. However, based on the human eye's observational characteristics, "condensing" the image and then shaking and dragging it can effectively detect MURA from the background. The potential energy function constructed above effectively simulates this observational characteristic of the human eye for low-contrast MURA. This step transforms the point mass set of the multi-scale channel into a set of potential energy vectors for the multi-scale channel.

[0067] S280, construct attention kernel function Among them, X is the potential energy vector of the multi-scale channel after the potential energy function conversion, and M is the state observation table that describes the subjective requirements of the human eye.

[0068] In one example, the potential energy function constructed above is used to characterize the degree of aggregation of local feature quantities, and needs to be further associated with the human eye attention characteristics. Let the attention kernel function be Where X is the potential energy vector set of the multi-scale channel obtained by converting the potential energy function constructed above; M is a state observation table that describes the subjective requirements of the human eye (such as specific visual inspection specifications for different product models). The state observation table represents subjective preferences such as perspective sensitivity, regional sensitivity, and morphological sensitivity in the form of state coefficients.

[0069] In one embodiment, the display panel defect detection method based on attention potential energy further includes: obtaining a preferred area in the target image associated with the panel to be detected; and automatically generating an associated state observation table based on the preferred area. The preferred area refers to an area that is set as a preference according to the detection requirements, and the preferred area is a partial area in the target image. The preferred area can be set manually and based on the detection requirements. For example, the preferred area can include but is not limited to one of the following: a key sensitive area, a negligible area, and a grayscale limit area, etc., and can be located at various positions in the target image, such as surrounding positions and a center position. The state observation table is a state observation table that describes the subjective requirements of the human eye (such as specific visual inspection specifications for panels to be detected for different product models); the state observation table characterizes subjective preferences such as visual sensitivity, regional sensitivity, and morphological sensitivity in the form of state coefficients. Table 1 is a schematic table of a state observation table provided in an embodiment of the present invention. As shown in Table 1, the values ​​in the first three rows and the first three columns in Table 1 represent the weight coefficients of the center position and the four sides and four corners. For example, the weight coefficient of the center position can be 1.00, the weight coefficient of the four corners can be 0.6, and the weight coefficient of the four sides can be 1.5; the values ​​in the first three rows and the 4th to 7th columns in Table 1 represent the number and width of the area to be detected. For example, the area numbers are 101, 102, 103, 104, 105 and 0.00, and the corresponding area widths are: 20.00, 20.00, 15.00, 15.00, 0.00 and 0.00 respectively; the values ​​in the 4th to 6th rows and the 1st to 7th columns in Table 1 represent the viewing angle sensitivity angle and the corresponding area number. For example, the viewing angle sensitivity angles are: 90.00, 45.00 and 10.00, and the area corresponding to 90.00 is 0.00. The domain numbers are 101, 102, 103, 104, 105 and 0.00 respectively, and the region numbers corresponding to 45.00 are 101, 102, 103, 104 and 0.00 respectively; and the region numbers corresponding to 10.00 are 101, 102, 0.00, 0.00, 0.00 and 0.00 respectively; the values ​​in rows 7-8 and columns 1-7 in Table 1 represent the defect sample type number, neighborhood major axis, neighborhood minor axis and quantity of the display panel. For example, the defect sample type number of the display panel is 11, and its neighborhood major axis, neighborhood minor axis and quantity are 15.00, 15.00 and 3 respectively, indicating that the defect sample type is dark spot interval; the defect sample type number of the display panel is 12, and its neighborhood major axis, neighborhood minor axis and quantity are 20.00, 60.00 and 2.00 respectively, indicating that the defect sample type is vertical MURA.

[0070] Table 1

[0071]

[0072]

[0073] S290. Apply the attention kernel function to the potential energy function to obtain the attention potential energy function.

[0074] In an embodiment, the attention kernel function can be applied to the potential energy function to obtain the attention potential energy function

[0075] S2100. Apply the attention potential energy function to the multi-scale channel potential energy vector set to obtain the attention potential energy vector set of the multi-scale channel.

[0076] In one embodiment, when a target image associated with a panel to be inspected includes singular points and singular regions, the display panel defect detection method based on attention potential energy further includes: obtaining a binary image mask of the singular points and singular regions in the target image;

[0077] Correspondingly, an attention potential energy function is constructed, and the attention potential energy vector set of the multi-scale channel is obtained based on the point mass volume set of the multi-scale channel, which also includes: determining the perturbation coefficients of the image pixels associated with the singular points and singular regions based on the binary image mask and assigning specific values; substituting the perturbation coefficients into the attention potential energy function to obtain the attention potential energy vector set of the multi-scale channel excluding the singular points and singular regions.

[0078] In one example, the singular points and singular regions refer to positions corresponding to external foreign objects in the external environment in the target image. The distribution and occupied areas of the singular points and singular regions on the target image are not fixed.

[0079] During the defect detection process of the panel to be inspected, it is inevitable that it will be affected by surface foreign matter such as dust in the environment. These external foreign matter are not real defects. For example, the arrow prints and labels on the polarizer of the display module also need to be ignored. If the above parts are not processed, after the above steps, a very strong attention potential energy response will be formed in the attention potential energy vector set, forming singular points and singular areas. This result itself is reasonable because the human eye itself responds strongly to these obvious features, but these are external interferences and need to be discarded in the detection results, but cannot be discarded or filtered out in the attention potential energy vector set that has been generated, because singular points and singular areas lead to strong attention responses that are far larger than their own area, and it is necessary to add singular point and singular area processing in the above processing steps.

[0080] After conventional image processing (filtering, dynamic threshold segmentation, morphological post-processing) is performed on the original image of the panel to be inspected (for example, the dust removal image and the label image), multiple binary image masks s0 to sn (dust removal mask, label mask, etc.) containing singular points and singular regions in the target image can be obtained. These image masks indicate the position and morphology of the singular points and singular regions in the target image. Furthermore, the perturbation coefficient C in the potential energy function of the above-mentioned point mass body can be assigned according to the singular positions corresponding to the binary image masks. Here, the singular points and singular regions are interference terms, so the perturbation coefficient C is assigned to 0, thereby obtaining the potential energy function that excludes the processing of singular points and singular regions: Correspondingly, we can obtain the potential energy vector set of the multi-scale channel excluding singular points and singular regions.

[0081] In one example, when the target image associated with the panel to be detected does not contain a singular position, the attention potential energy function is constructed based on the attention kernel function and the potential energy function, which is recorded as Among them, att(i,j) means that the pixel coordinate (i,j) is within the range of the field radius. The attention potential vector within .

[0082] In one example, when the target image associated with the panel to be inspected contains singular positions, multiple binary image masks s0~sn (dust removal masks, label masks, etc.) containing singular points and singular regions in the target image can be obtained. These image masks indicate the positions and shapes of the singular points and singular regions in the target image. Furthermore, the perturbation coefficient C in the potential energy function of the above-mentioned point mass body can be assigned according to the singular positions corresponding to the binary image masks. Here, the singular points and singular regions are interference terms, so the perturbation coefficient C is assigned to 0. The attention potential energy function is constructed based on the attention kernel function and the potential energy function, which is recorded as Among them, att'(i,j) means that the pixel coordinate (i,j) is within the range of the field radius. The attention potential energy vector within the image is obtained by excluding the above-mentioned dust, foreign matter, vector prints, labels and other image masks.

[0083] It can be understood that the attention kernel function is applied to the potential energy function constructed above, that is, the potential energy set vector of the multi-scale channel is converted into the attention potential energy vector set of the multi-scale channel. This attention potential energy vector set uses the original pixels of the image as the basic material, and through the potential energy function and the attention kernel function, it integrates the subjective preference factors of the human eye, and has a strong alignment with the inspection results of the human eye.

[0084] S2110. Perform adaptive threshold processing on the attention potential energy vector set of the multi-scale channel to obtain the attention potential energy value and attention potential energy range of the multi-scale channel.

[0085] Furthermore, the acquired attention potential energy vector set of the multi-scale channel is subjected to adaptive threshold processing (note that the threshold here is the attention potential energy threshold, not the image grayscale threshold) to obtain the specific attention potential energy value (also called the observation intensity value) and the corresponding attention potential energy range (also called the observation field value). Note that this observation intensity value can be used as the final evaluation value.

[0086] S2120: Perform a local inverse transformation on the attention potential energy value and the attention potential energy range of the multi-scale channel to obtain the physical intensity value and the physical intensity range of the multi-scale channel.

[0087] Since the observed field value is generally quite different from the physical value of the actual object defect, it is necessary to perform a local inverse transformation on the aforementioned determination process of the attention potential energy value and the attention potential energy range to obtain the physical intensity value and the physical intensity range in the image, and then obtain the true physical value of the defect.

[0088] S2130: Merge the physical intensity values ​​of the multi-scale channels at the same coordinates, and obtain a defect detection result of the panel to be inspected based on the physical intensity range of the multi-scale channels.

[0089] In an embodiment, the attention potential energy values ​​(i.e., physical intensity values) of the multi-scale channels of the same coordinates can be aggregated and merged (i.e., accumulated), and the defect detection result of the panel to be inspected can be given as qualified or unqualified based on the set threshold value (observation intensity value and observation field value).

[0090] The technical solution of this embodiment abandons the classic image processing category and the traditional deep learning category through a detection algorithm that can be strongly aligned with the subjective senses of the human eye. By constructing an attention potential energy function to simulate the characteristics of the human eye, the detection results are highly aligned with the human eye.

[0091] In one embodiment, Figure 3 This is a configuration flow chart of a state observation table provided by an embodiment of the present invention, such as Figure 3 As shown in the figure, the configuration process of the state observation table includes the following steps:

[0092] S310 , obtaining a target image of a limit sample of a panel to be inspected.

[0093] Generally, the number of target images of the limit sample of the panel to be inspected may be 3 to 5.

[0094] S320: Determine a preferred area in the target image.

[0095] Depending on the model of the panel to be inspected, manual preference settings may be required, such as key sensitive areas, ignored areas, grayscale limits, etc.

[0096] S330: Generate a state observation table corresponding to the preferred area.

[0097] In an embodiment, the program automatically generates a status observation table according to the settings of these preference areas.

[0098] In one embodiment, Figure 4 This is a flowchart of an implementation of a display panel defect detection based on attention potential provided by an embodiment of the present invention. Based on the above embodiment, this embodiment is a preferred embodiment and takes the three physical scales including micrometers, millimeters and centimeters as examples to illustrate the display panel defect detection process based on attention potential. Figure 4 As shown, the defect detection process of the display panel based on attention potential in this embodiment includes the following steps:

[0099] Step 1: Obtain the original image of the panel to be inspected.

[0100] Step 2: perform image preprocessing on the original image to obtain the target image.

[0101] Step 3: Preprocess the target image to obtain the corresponding image mask.

[0102] In an embodiment, a target image including surface dust removal, labels, etc. for mask processing is acquired; the input image including dust removal, labels, etc. is preprocessed to generate a corresponding image mask.

[0103] Step 4: perform scale separation on the target image to obtain an image set of multi-scale channels.

[0104] In the embodiment, the scale set conversion of the three channels is beneficial to improving the timeliness of subsequent processing steps; it is noted that this step simultaneously performs scale separation on the above-mentioned image mask.

[0105] Step 5: Convert the image set of the multi-scale channels into a point mass volume set of the corresponding multi-scale channels.

[0106] In one example, image sets of three scale channels are converted into corresponding point mass volume sets, namely point mass volume set 1, point mass volume set 2, and point mass volume set 3.

[0107] In an embodiment, the image set of the multi-scale channel is converted into a point mass volume set of the multi-scale channel: g(i,j)=h(i,j)*I(i,j), where h(i,j) is a quality weight coefficient that varies according to position. In this example, the weight coefficient h(i,j) is taken as a constant h0; note that this step also converts the corresponding image mask mentioned above.

[0108] Step 6: Obtain the state observation table corresponding to each image.

[0109] Step 7, potential energy processing.

[0110] Perform parallel potential energy processing on point mass set 1, point mass set 2 and point mass set 3 to obtain corresponding potential energy vector sets. In this embodiment, the potential energy functions of three different sizes of r0, r1 and r2 are described by the above method. Perform potential energy processing and parallel processing of three channels r0, r1, and r2 to speed up the calculation.

[0111] Step 8: Construct the attention kernel function.

[0112] Based on the potential energy vector set, the attention kernel functions of point mass set 1, point mass set 2 and point mass set 3 are constructed in parallel.

[0113] Step 9: Attention potential energy processing.

[0114] An attention potential energy function is constructed based on the attention kernel function and the potential energy function. Based on the attention potential energy function, the potential energy vector set of each scale channel is converted into the attention potential energy vector set of multi-scale channels in parallel.

[0115] Step 10: Processing of singular points and singular regions.

[0116] This step requires using the image mask generated above to process singular points and singular areas, and adjusting parameters to ensure that the detection results are well aligned with the human eye observation results.

[0117] Step 11, inverse transform.

[0118] The acquired attention potential energy vector set is subjected to adaptive threshold processing to obtain a specific attention potential energy value (observation intensity value); at the same time, a local inverse transformation is performed to obtain the domain range value in the image, and then the real physical value of the defect is obtained.

[0119] Step 12: Output the defect detection result of the panel to be inspected.

[0120] Figure 5 This is a schematic diagram showing an original image with a strong defect and its attention potential energy vector diagram provided by an embodiment of the present invention; Figure 6 This is a schematic diagram showing an original image of a dirt defect and its attention potential energy vector diagram provided by an embodiment of the present invention; Figure 7 8 is a schematic diagram of the display of an original image of a vertical MURA defect and its attention potential energy vector diagram provided by an embodiment of the present invention; 8 is a schematic diagram of the display of an original image of a black GAP defect and its attention potential energy vector diagram provided by an embodiment of the present invention.

[0121] like Figure 5 As shown in the figure, the defect can be clearly seen in the original image of the strong defect (left figure), and in its attention potential energy vector diagram (right figure), the red color is used to indicate that the stronger the contrast of this area, the more obvious the corresponding defect; Figure 6 、 7 As shown in Figure 8, the original image contains a weak defect (left image). The defect cannot be clearly seen in the original image, but in its corresponding attention potential energy vector diagram (right image), it can be seen Figure 6 The right image in the figure contains relatively weak contrast colors such as orange, and the corresponding defects are not very obvious; Figure 7 The right image in the figure contains a small amount of colors with slightly stronger contrast, such as red, and relatively weaker contrast, such as yellow, and the corresponding defects are not very obvious; Figure 8 The right image in contains only the least contrasting colors, such as cyan, where the defect is least noticeable.

[0122] In one embodiment, Figure 9 FIG is a schematic diagram of the structure of a display panel defect detection device based on attention potential energy provided by an embodiment of the present invention. Figure 9 As shown, the device includes: a first acquisition module 910 , a first conversion module 920 , a second conversion module 930 and a detection module 940 .

[0123] The first acquisition module 910 is configured to acquire an image set of multi-scale channels associated with the panel to be inspected;

[0124] A first conversion module 920 is configured to assign a quality value to each image pixel in the image set of the multi-scale channel to obtain a point quality volume set of the multi-scale channel;

[0125] A second conversion module 930 is used to construct an attention potential energy function and obtain an attention potential energy vector set of the multi-scale channel based on the point mass volume set of the multi-scale channel;

[0126] The detection module 940 is configured to obtain a defect detection result of the panel to be inspected based on the attention potential energy value and the attention potential energy range associated with the attention potential energy vector set of the multi-scale channel.

[0127] The technical solution of this embodiment assigns a mass value to each image pixel contained in each image in the image set of the multi-scale channel associated with the panel to be inspected, thereby obtaining a point mass volume set of the multi-scale channel, and simulating the characteristics of the human eye by constructing an attention potential energy function. The attention potential energy vector set of the multi-scale channel is converted based on the point mass volume set of the multi-scale channel, and the defect detection result of the panel to be inspected is obtained based on the attention potential energy value associated with the attention potential energy vector set of the multi-scale channel, so that the defect detection result has a highly aligned relationship with the human eye, thereby reducing the distortion of the detection result.

[0128] In one embodiment, the first acquisition module 910 includes:

[0129] A first acquisition unit, configured to acquire an original image associated with the panel to be inspected;

[0130] An image preprocessing unit, configured to perform image preprocessing on the original image to obtain a related target image;

[0131] The scale separation unit is used to perform a scale separation operation on the target image to obtain an image set of multi-scale channels.

[0132] In one embodiment, the first conversion module 920 includes:

[0133] A second acquisition unit is used to acquire each image pixel of each image in the image set of the multi-scale channel;

[0134] The first determining unit is configured to multiply each image pixel by an associated quality weight coefficient to obtain a point mass volume set of a multi-scale channel.

[0135] In one embodiment, the second conversion module 930 includes:

[0136] The first function construction unit is used to construct the potential energy function of a point mass body Among them, G(i,j) represents the point mass body at the pixel coordinate (i,j) within the field radius range The potential energy in the inner space; k represents the deviation length between the x-axis and i; l represents the deviation length between the y-axis and j; c represents the perturbation coefficient;

[0137] A first conversion unit is used to apply a potential energy function to a point mass set of the multi-scale channel to obtain a potential energy vector set of the multi-scale channel;

[0138] The second function construction unit is used to construct the attention kernel function Where X is the potential energy vector of the multi-scale channel after the potential energy function conversion, and M is the state observation table that describes the state that meets the subjective requirements of the human eye;

[0139] The third function construction unit is used to apply the attention kernel function to the potential energy function to obtain the attention potential energy function

[0140] The second conversion unit is used to apply the attention potential energy function to the multi-scale channel potential energy vector set to obtain the attention potential energy vector set of the multi-scale channel.

[0141] In one embodiment, the detection module 940 includes:

[0142] A processing unit, configured to perform adaptive threshold processing on the attention potential energy vector set of the multi-scale channel to obtain the attention potential energy value and attention potential energy range of the multi-scale channel;

[0143] An inverse transformation unit, configured to perform a local inverse transformation on the attention potential energy value and the attention potential energy range of the multi-scale channel to obtain the physical intensity value and the physical intensity range of the multi-scale channel;

[0144] The merging unit is used to merge the physical intensity values ​​of the multi-scale channels of the same coordinates, and obtain the defect detection result of the panel to be detected based on the physical intensity range of the multi-scale channels.

[0145] In one embodiment, when a target image associated with a panel to be inspected contains singular points and singular regions, the display panel defect detection apparatus based on attention potential energy further includes:

[0146] A second acquisition module is used to obtain a binary image mask of a singular position in the target image;

[0147] Correspondingly, the second conversion module 930 further includes:

[0148] A second determining unit is configured to determine a disturbance coefficient of an image pixel associated with a singular position based on the binary image mask and assign a specific value to the disturbance coefficient;

[0149] The third conversion unit is used to substitute the perturbation coefficient into the attention potential energy function to obtain a set of attention potential energy vectors of the multi-scale channel excluding singular points and singular regions.

[0150] In one embodiment, the display panel defect detection device based on attention potential energy further includes:

[0151] A third acquisition module is used to acquire a preferred area in a target image associated with the panel to be inspected;

[0152] A generation module is used to automatically generate an associated state observation table based on the preference area.

[0153] The display panel defect detection device based on attention potential provided by an embodiment of the present invention can execute the display panel defect detection method based on attention potential provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0154] In one embodiment, Figure 10 This is a structural block diagram of an electronic device provided by an embodiment of the present invention. Figure 10 , a schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0155] like Figure 10 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., which is communicatively connected to the at least one processor 11. The memory stores a computer program that can be executed by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. Various programs and data required for the operation of the electronic device 10 can also be stored in the RAM 13. The processor 11, ROM 12, and RAM 13 are connected to each other via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0156] Multiple components in the electronic device 10 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0157] The processor 11 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the display panel defect detection method based on attention potential.

[0158] In some embodiments, the display panel defect detection method based on attention potential can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the display panel defect detection method based on attention potential described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to execute the display panel defect detection method based on attention potential in any other appropriate manner (for example, by means of firmware).

[0159] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0160] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0161] In the context of the present invention, computer-readable storage media can be tangible media that can contain or store a computer program for use with an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. Computer-readable storage media can include but are not limited to electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, computer-readable storage media can be machine-readable signal media. More specific examples of machine-readable storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0162] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0163] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0164] A computing system may include clients and servers. The clients and servers are typically remote from each other and typically interact via a communication network. This client-server relationship arises through computer programs running on the respective computers, creating a client-server relationship. The server may be a cloud server, also known as a cloud computing server or cloud host. This server is a hosting product within the cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosting and VPS services.

[0165] An embodiment of the present invention also provides a computer program product, including a computer program, which, when executed by a processor, can implement the display panel defect detection method based on attention potential as provided in any embodiment of the present application.

[0166] The computer program product, during implementation, may be written in one or more programming languages ​​or a combination thereof, for performing the operations of the present application, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0167] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in the present invention can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of the present invention can be achieved. This is not limited herein.

[0168] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A display panel defect detection method based on attention potential, characterized in that: include: Obtain an image set of multi-scale channels associated with the panel to be detected; Assigning a mass value to each image pixel in the image set of the multi-scale channel to obtain a point mass volume set of the multi-scale channel; Constructing an attention potential energy function, and obtaining an attention potential energy vector set of the multi-scale channel according to the point mass volume set of the multi-scale channel; Obtaining a defect detection result of the panel to be inspected based on the attention potential energy value and the attention potential energy range associated with the attention potential energy vector set of the multi-scale channel; The attention potential energy function is constructed, and the attention potential energy vector set of the multi-scale channel is obtained according to the point mass volume set of the multi-scale channel, including: Constructing the potential energy function of a point mass ;in, Represents pixel coordinates The point mass at is within the radius of the field internal potential energy; Indicates that in the x-axis direction The length of deviation between Indicates that in the y-axis direction The length of deviation between represents the disturbance coefficient; is the pixel coordinate point mass; Applying the potential energy function to the point mass set of the multi-scale channel to obtain a potential energy vector set of the multi-scale channel; Constructing the attention kernel function , where X is the potential energy vector of the multi-scale channel after the potential energy function is converted, and M is a state observation table that describes the state that meets the subjective requirements of the human eye; Apply the attention kernel function to the potential energy function to obtain the attention potential energy function ; The attention potential energy function is applied to the multi-scale channel potential energy vector set to obtain the attention potential energy vector set of the multi-scale channel.

2. The method according to claim 1, characterized in that The step of obtaining an image set of multi-scale channels associated with the panel to be detected includes: Obtaining the original image associated with the panel to be inspected; Performing image preprocessing on the original image to obtain a related target image; A scale separation operation is performed on the target image to obtain an image set of multi-scale channels.

3. The method according to claim 1, characterized in that The step of assigning a quality value to each image pixel in the image set of the multi-scale channel to obtain a point mass volume set of the multi-scale channel includes: Obtain each image pixel of each image in the image set of the multi-scale channel; Each of the image pixels is multiplied by the associated quality weight coefficient to obtain a point mass volume set of a multi-scale channel.

4. The method according to claim 1, wherein Obtaining a defect detection result of the panel to be inspected based on the attention potential energy value and the attention potential energy range associated with the attention potential energy vector set of the multi-scale channel, including: Performing adaptive threshold processing on the attention potential energy vector set of the multi-scale channel to obtain the attention potential energy value and attention potential energy range of the multi-scale channel; Performing a local inverse transformation on the attention potential energy value and the attention potential energy range of the multi-scale channel to obtain the physical intensity value and the physical intensity range of the multi-scale channel; The physical intensity values ​​of the multi-scale channels at the same coordinates are merged, and a defect detection result of the panel to be inspected is obtained based on the physical intensity range of the multi-scale channels.

5. The method according to claim 1, wherein In a case where the target image associated with the panel to be inspected contains singular points and singular areas, the method further includes: Obtaining a binary image mask of singular points and singular regions in the target image; Correspondingly, the method of constructing an attention potential energy function and obtaining an attention potential energy vector set of a multi-scale channel based on the point mass volume set of the multi-scale channel further includes: Determine the disturbance coefficients of the image pixels associated with the singular points and the singular regions based on the binary image mask and assign specific values ​​to them; Substituting the perturbation coefficient into the attention potential energy function, a set of attention potential energy vectors of the multi-scale channel excluding the singular points and singular regions is obtained.

6. The method according to any one of claims 1 to 5, characterized in that The method further comprises: Acquiring a preferred area in a target image associated with the panel to be inspected; An associated state observation table is automatically generated based on the preference area.

7. A display panel defect detection device based on attention potential energy, characterized in that: include: A first acquisition module is used to acquire an image set of multi-scale channels associated with the panel to be detected; A first conversion module is configured to assign a quality value to each image pixel in the image set of the multi-scale channel to obtain a point mass volume set of the multi-scale channel; A second conversion module is used to construct an attention potential energy function and obtain an attention potential energy vector set of the multi-scale channel according to the point mass volume set of the multi-scale channel; a detection module, configured to obtain a defect detection result of the panel to be inspected based on the attention potential energy value and the attention potential energy range associated with the attention potential energy vector set of the multi-scale channel; Constructing an attention potential energy function, and obtaining an attention potential energy vector set of the multi-scale channel according to the point mass volume set of the multi-scale channel, including: Constructing the potential energy function of a point mass ;in, Represents pixel coordinates The point mass at the area is within the radius of the field internal potential energy; Indicates that in the x-axis direction The length of deviation between Indicates that in the y-axis direction The length of deviation between represents the disturbance coefficient; is the pixel coordinate point mass; Applying the potential energy function to the point mass set of the multi-scale channel to obtain a potential energy vector set of the multi-scale channel; Constructing the attention kernel function , where X is the potential energy vector of the multi-scale channel after the potential energy function is converted, and M is a state observation table that describes the state that meets the subjective requirements of the human eye; Apply the attention kernel function to the potential energy function to obtain the attention potential energy function ; The attention potential energy function is applied to the multi-scale channel potential energy vector set to obtain the attention potential energy vector set of the multi-scale channel.

8. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the display panel defect detection method based on attention potential energy according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the display panel defect detection method based on attention potential energy according to any one of claims 1 to 6 when executed.

10. A computer program product, characterized in that The computer program product comprises a computer program, which, when executed by a processor, implements the display panel defect detection method based on attention potential according to any one of claims 1 to 6.

Images