Method and system for enhancing Mura defect image in OLED display panel

Abstract

The invention provides a method and system for enhancing a Mura defect image in an OLED display panel, and relates to the field of defect detection and quality control, and the method comprises the steps: obtaining an original image of the OLED display panel, the original image having a Mura defect; performing histogram image conversion and pseudo-infrared image conversion on the original image by adopting a dual-channel preprocessing strategy to obtain two complementary histogram enhanced images and pseudo-infrared images; wherein the histogram image conversion process comprises color image graying, local histogram statistics, equalization mapping and smooth fusion, and the pseudo infrared image conversion process comprises contrast and brightness adjustment and pseudo infrared color mapping; and performing three-mode feature fusion on the original image, the histogram enhanced image and the pseudo-infrared image to obtain a defect integrated image. According to the method, a complementary information fusion image preprocessing mode is adopted, histogram equalization and pseudo-thermal imaging transformation are combined, and the visibility of low-contrast defects can be enhanced.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of defect detection and quality control, in particular to a Mura defect image enhancement method and system in an OLED display panel. BACKGROUND

[0002] OLED display panels have been widely used in high-end consumer electronics due to their excellent display performance, but Mura defects in OLED display panels are too weak to be detected, which has seriously affected product quality and production cost.

[0003] Mura defects are usually manifested as weakly uneven areas in brightness, and the difference with the surrounding background is often only 3-5%, which is almost indistinguishable in the original captured image. More complex is that different types of Mura defects exhibit different visual appearances, and the appearance of the same type of Mura defect in different production batches can also have great differences, which makes it difficult for feature extraction methods relying on a single image domain to fully capture the defect characteristics. In the actual production environment, additional noise and artifacts are introduced due to light fluctuations, camera parameter adjustments or dust pollution, further reducing image quality; at the same time, the ratio of defect-free samples to defective samples is severely imbalanced, and the distribution of different types of defects is also extremely uneven, which leads to missed detection, false detection or class bias of detection models based on conventional preprocessing or simple enhancement.

[0004] The commonly used image enhancement methods in the prior art, such as brightness adjustment and contrast stretching, often have little effect when faced with Mura defects of low contrast and weak defects, and can also introduce new noise, and rely on static preprocessing parameters, which are difficult to adapt to the diversity of different batches of panels. At the same time, schemes that simply process in the gray domain or frequency domain cannot take into account the multi-dimensional features of different defects, and manual annotation with high annotation cost further restricts the acquisition of effective training data. Therefore, to significantly improve the accuracy and stability of defect detection, a special image enhancement method must be designed at the data level that can both enhance Mura features and suppress noise and background interference, and reduce dependence on expensive hardware and a large amount of labeled data. SUMMARY

[0005] The present application provides a Mura defect image enhancement method and system in an OLED display panel, which cannot improve the accuracy and stability of Mura defect detection through software technology in the prior art.

[0006] In a first aspect, a Mura defect image enhancement method in an OLED display panel is provided in embodiments of the present application, and the method includes the following processes:

[0007] An original image of the OLED display panel is acquired, wherein the original image has a Mura defect;

[0008] A histogram image conversion and a pseudo-infrared image conversion are performed on the original image by using a double-channel preprocessing strategy to obtain two complementary histogram enhanced images and a pseudo-infrared image; wherein the histogram image conversion process comprises color image gray scaling, local histogram statistics, equalization mapping and smoothing fusion, and the pseudo-infrared image conversion process comprises contrast and brightness adjustment and pseudo-infrared color mapping.

[0009] Three-mode feature fusion is performed on the original image, the histogram enhanced image and the pseudo-infrared image to obtain a defect integrated image.

[0010] In the above embodiments, the present application discards the dependence on additional imaging hardware such as a depth sensor or a thermal imager, and instead uses a pseudo-infrared image and a histogram enhanced image and an original defect image to simulate multi-modal input, without the need for additional thermal imagers or depth sensors, without increasing production costs, without complicating the system, and from three complementary angles of contrast, pseudo-thermal color and original texture, fully strengthening the low-contrast and uneven Mura defect area, thereby more accurately positioning the defect position. The algorithm structure is simple, real-time, can be quickly fed back online, and can meet the immediate correction needs.

[0011] As some optional embodiments of the present application, the process of the histogram image conversion is as follows:

[0012] Color image gray scaling is performed on the original image by using a weighted average algorithm to obtain a single-channel gray image;

[0013] The single-channel gray image is divided into a plurality of small blocks, and the probability distribution of the gray value in each small block is calculated to realize local histogram statistics;

[0014] Based on the local histogram statistics result, the gray value of each small block is mapped and expanded to the full gray level range to realize local equalization mapping;

[0015] Based on the local equalization mapping result, the plurality of small blocks are smoothed and fused to generate an overall histogram enhanced image.

[0016] In the above embodiments, the present application divides the gray image, performs local histogram statistics, local equalization and smoothing fusion, which is simpler in parameter tuning, without the need for re-distributing the histogram at the sub-block level, and improves the universality and stability of the algorithm.

[0017] As some optional embodiments of the present application, the process of the pseudo-infrared image conversion is as follows:

[0018] The original image is subjected to contrast and brightness adjustment algorithm to obtain an adjusted image with local gray scale variation;

[0019] The adjusted image with local gray scale variation is subjected to JET color map mapping to map the gray scale value to a gradual color space to obtain a pseudo-infrared image.

[0020] In the above embodiment, the application adopts a two-channel preprocessing strategy to take the two processed images as two-channel input, retains the original brightness information while introducing color enhancement representation, forms a complementary multi-domain feature space, significantly improves the visibility of low-contrast Mura defects, and converts the brightness abnormalities that are difficult to detect by naked eyes into easily identifiable feature representation.

[0021] As some optional embodiments of the present application, the process of the three-modal feature fusion is as follows:

[0022] The original image, the histogram enhanced image and the pseudo-infrared image are subjected to size normalization to ensure that the spatial resolution of the images is the same;

[0023] The original image, the histogram enhanced image and the pseudo-infrared image with the same resolution are subjected to channel splicing in the channel dimension to obtain a multi-channel fusion feature image;

[0024] The multi-channel fusion feature image is subjected to dimension reduction processing to obtain a defect integrated image.

[0025] In the above embodiment, the three-modal feature fusion generates a rich feature image that takes into account details and the whole, laying a data foundation for Mura defect detection.

[0026] As some optional embodiments of the present application, the calculation formula of the weighted average algorithm is:

[0027] r j = 0.299 * R + 0.587 * G + 0.114 * B

[0028] wherein r j represents the gray scale value of the pixel position of the original image, and RGB represents the three primary colors of the pixel position of the original image.

[0029] In the above embodiment, the application adopts the weighted average algorithm to convert the original image from a color image to a gray scale image, and enables the gray scale image to maintain the corresponding texture features.

[0030] As some optional embodiments of the present application, the calculation formula of the probability distribution of the gray scale value in the small block is:

[0031]

[0032] Where p(r) j ) represents the probability distribution of grayscale values ​​within each small block, n j This indicates that the grayscale value within the small block is r. j The number of pixels, where n represents the total number of pixels within the block.

[0033] In the above embodiments, the present invention performs grayscale probability distribution analysis on each small block, and then performs local histogram statistics on each small block, which can improve the overall visual effect of the image.

[0034] As one of the optional embodiments of this application, the calculation formula for the smooth fusion of the small blocks is as follows:

[0035]

[0036] in, This represents a histogram-enhanced image, where L represents the total number of gray levels, r represents the gray value of a pixel in the original image, and p(r) j ) represents the probability distribution of grayscale values ​​within each small block.

[0037] In the above embodiments, the present invention performs smooth fusion on the small blocks after local histogram statistics and local equalization mapping. The smooth fusion process ensures that the gray values ​​of each small block are consistent in the global range, avoiding overall incoordination caused by excessive or insufficient local enhancement, so that the image can maintain global tone and brightness balance while enhancing local details.

[0038] As some optional embodiments of this application, the calculation formula of the contrast and brightness adjustment algorithm is as follows:

[0039]

[0040] in, This indicates adjusting the image, where I(x,y) represents the grayscale intensity at the pixel position (x,y). β represents the contrast adjustment factor, β represents the brightness compensation parameter, and the CLIP(·) function ensures that the pixel values ​​for contrast and brightness adjustment remain within the effective range.

[0041] In the above embodiments, the contrast and brightness adjustment algorithm of the present invention processes the original image, which can enhance image details and recognizability, and achieve flexible control of local contrast and brightness.

[0042] In a second aspect, the present invention provides an image enhancement system for Mura defects in an OLED display panel, the system comprising:

[0043] An original image acquisition unit is configured to acquire an original image of the OLED display panel, wherein the original image has a Mura defect;

[0044] A dual-channel preprocessing unit is configured to perform histogram image conversion and pseudo-infrared image conversion on the original image by using a dual-channel preprocessing strategy to obtain two complementary histogram enhanced images and a pseudo-infrared image; wherein the histogram image conversion process includes color image graying, local histogram statistics, equalization mapping and smoothing fusion, and the pseudo-infrared image conversion process includes contrast and brightness adjustment and pseudo-infrared color mapping.

[0045] A three-modal feature fusion unit is configured to perform three-modal feature fusion on the original image, the histogram enhanced image and the pseudo-infrared image to obtain a defect integrated image.

[0046] In a third aspect, the present application provides a computer device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the Mura defect image enhancement method in the OLED display panel when executing the computer program.

[0047] In a fourth aspect, the present application provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the Mura defect image enhancement method in the OLED display panel.

[0048] The present application has the following advantages:

[0049] 1. The present application replaces expensive infrared thermal imagers, depth sensors and other hardware with three modalities of original images, histogram enhanced images and pseudo-infrared images, and simulates multi-source information by software to greatly reduce system procurement and maintenance costs.

[0050] 2. The present application adopts an image preprocessing method of complementary information fusion, combines histogram equalization and pseudo-thermal imaging transformation, and can enhance the visibility of low-contrast defects. BRIEF DESCRIPTION OF DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can also be obtained from these drawings without creative labor.

[0052] Figure 1is a flowchart of the Mura defect image enhancement method according to an embodiment of the present application;

[0053] Figure 2 is a three-modality image contrast schematic diagram according to an embodiment of the present application. DETAILED DESCRIPTION

[0054] It should be understood that the specific embodiments described herein are merely intended to explain the present application and are not intended to limit the present application.

[0055] The defect detection system of the existing OLED display panel often relies on manual processing or complex algorithm processing, which not only increases the production cost, but also leads to complex system structure and high maintenance difficulty. In addition, a single original image preprocessing means is difficult to fully excavate the low-contrast Mura defect characteristics, so that the subsequent system is easy to be disturbed by noise and the positioning accuracy is limited. In view of the above shortcomings, the present application provides a Mura defect image enhancement method and system for OLED display panel, which can generate more discriminative sample data without additional hardware investment and increasing system complexity, which is helpful for more accurate positioning and identification of Mura defects, thereby improving product yield.

[0056] Please refer to Figure 1 , Figure 1 is a flowchart of the Mura defect image enhancement method, and the method flow is as follows:

[0057] Step 1: Obtain the original image of the OLED display panel.

[0058] In the embodiment of the present application, in the production process of the OLED display panel, the high-quality original image is obtained through the online monitoring system, and the real-time monitoring of the Mura defect is realized. According to the feedback of the online monitoring system, the relevant Mura defect data and the processing process data corresponding to the defective product are collected in time. Specifically, the relevant data collected by the online monitoring system includes but is not limited to production history data, intermediate optical detection data, AOI automatic optical sampling data and final display effect test data, etc.

[0059] In particular, the Mura defect data is not limited to the luminance non-uniformity value or the label of whether there is a bright spot (i.e., the OLED display panel is qualified or unqualified) detected by the final display effect of the OLED display panel, but also can be combined with the OLED display panel manufacturing business logic to flexibly define the Mura defect type according to the parameter abnormality (such as local luminance deviation from the standard value, color consistency abnormality, etc.) in the production monitoring process, so as to realize early discovery and disposal of the luminance abnormality of the OLED display panel. Therefore, the related Mura defect data can be the final optical detection result data of the OLED display panel (such as the luminance uniformity index or the category label of whether the panel has a visible bright spot, etc.), or the monitoring data in the intermediate process (such as an abnormally high light-emitting layer thickness parameter or TFT drive voltage abnormal data, etc.).

[0060] Specifically, the field information provided by the Mura defect data includes but is not limited to: 1. ID of the defective panel; 2. Process site of the Mura defect; 3. Number of defective panels; 4. Total input amount of panels in the same batch; 5. Time of obtaining the Mura defect; 6. Type of the Mura defect; and 7. Position information of the Mura defect.

[0061] Specifically, the processing process data corresponding to the defective product includes but is not limited to: recording the acquisition source (such as camera ID, station number), resolution, image format, image acquisition state, whether there is a defect label, exposure parameter, and ambient light value, etc. for each original image.

[0062] Step 2: Adopting a double-channel preprocessing strategy to perform histogram image conversion and pseudo-infrared image conversion on the original image to obtain two complementary histogram enhanced images and a pseudo-infrared image.

[0063] Specifically, the histogram image conversion process includes color image grayscale, local histogram statistics, equalization mapping, and smoothing fusion, and the pseudo-infrared image conversion process includes contrast and brightness adjustment and pseudo-infrared color mapping.

[0064] In the embodiment of the present application, the flow of the histogram image conversion is as follows:

[0065] Step A2.1: Adopting a weighted average algorithm to perform color image grayscale on the original image to obtain a single-channel grayscale image.

[0066] Specifically, the calculation formula of the weighted average algorithm is:

[0067] r j =0.299×R+0.587×G+0.114×B

[0068] Wherein, r jGray value representing different pixel positions of the original image, RGB represents three primary colors of different pixel positions of the original image.

[0069] Step A2.2: The single-channel gray image is divided into several small blocks, and the probability distribution of the gray value in each small block is calculated to realize local histogram statistics.

[0070] Specifically, the calculation formula of the probability distribution of the gray value in the small block is:

[0071]

[0072] Wherein, p(r j ) represents the probability distribution of the gray value in each small block, n j represents the number of pixels with gray value r j in the small block, and n represents the total number of pixels in the small block.

[0073] Step A2.3: Based on the local histogram statistics result, the gray value of each small block is mapped and expanded to the full gray level range to realize local equalization mapping.

[0074] Step A2.4: Based on the local equalization mapping result, the several small blocks are smoothly fused to generate an overall histogram enhancement image, and the histogram enhancement image is saved in a unified format, and the file name records the batch and preprocessing parameters.

[0075] Specifically, the calculation formula of the smooth fusion of the small block is:

[0076]

[0077] Wherein, represents the histogram enhancement image, L represents the total number of gray levels, r represents the gray value of the pixel in the original image, and p(r j ) represents the probability distribution of the gray value in each small block.

[0078] In the embodiment of the application, the process of the pseudo-infrared image conversion is as follows:

[0079] Step B2.1: The contrast and brightness adjustment algorithm is used to adjust the contrast and brightness of the original image to obtain an adjusted image with local gray value change.

[0080] Specifically, the calculation formula of the contrast and brightness adjustment algorithm is:

[0081]

[0082] Wherein, represents the adjusted image, I(x,y) represents the gray intensity of the display pixel position (x,y), represents a contrast adjustment factor, β represents a brightness compensation parameter, and the CLIP(·) function ensures that the pixel values of the contrast and brightness adjustment remain in the valid range; in the embodiment of the present application, β is set to β = -255, emphasizing the defect area by local gray scale change.

[0083] Step B2.2: Perform JET color map mapping on the local gray scale change adjustment image using a predefined mapping rule to map the gray scale value to a gradient color space to obtain a pseudo-infrared image.

[0084] Wherein, the predefined mapping rule: low gray value is mapped to cold color (blue and purple), and high gray value is mapped to warm color (red and yellow).

[0085] Specifically, the gray scale value is mapped to the gradient color space by the applyColorMap function of OpenCV, and then the pseudo-infrared image is obtained; and the pseudo-infrared image is saved according to a unified file name, and the adjustment parameters and mapping rule settings are recorded to ensure the reproducibility and comparison.

[0086] Specifically, the dual-channel preprocessing strategy adopted in the embodiment of the present application can use the two preprocessed images as dual-channel input, while retaining the original brightness information and introducing color enhancement representation, forming a complementary multi-domain feature space, which can significantly improve the visibility of Mura defects and convert the brightness abnormalities that are difficult to detect by the naked eye into easily identifiable feature representation.

[0087] Step 3: Perform three-modal feature fusion on the original image, the histogram enhancement image and the pseudo-infrared image to obtain a defect integrated image.

[0088] In the embodiment of the present application, the process of three-modal feature fusion is as follows:

[0089] Step 3.1: Size normalization is performed on the original image, the histogram enhancement image and the pseudo-infrared image to ensure that the spatial resolution (height H and width W) of the images is the same, so as to facilitate subsequent splicing and fusion.

[0090] Step 3.2: The original image, the histogram enhancement image and the pseudo-infrared image with the same resolution are spliced in the channel dimension to obtain a multi-channel fusion feature image.

[0091] Specifically, in PyTorch, F cat = Concat(R,T,D) is called for splicing; wherein, R represents the original image, D represents the histogram enhancement image, T represents the pseudo-infrared image, Concat(·) represents a concatenation operation along the channel dimension, and F cat represents a multi-channel fusion feature image.

[0092] Step 3.3: Dimensionality reduction processing is performed on the multi-channel fusion feature image to obtain a defect integrated image.

[0093] In the embodiment of the present application, if each input is a C channel, the fusion feature image is a 3C channel, which simultaneously contains key information under three viewing angles. Specifically, after completing channel splicing, the channel number is usually reduced to a proper size in the next 3x3 convolution layer, so as to be used by subsequent multi-scale feature extraction and final classification and positioning network. The defect integrated image is a rich feature image that takes into account details and the whole, which lays a data foundation for Mura defect detection.

[0094] Next, a specific implementation form of the embodiment of the present application is given for Mura defects in the production process of an OLED display panel. It should be noted that this implementation form is only an implementation form of the embodiment of the present application, but not all implementation forms of the embodiment of the present application. Meanwhile, the Mura defect image enhancement method proposed in the embodiment of the present application is also not limited to the field of OLED display panel manufacturing.

[0095] S10: Obtain an original image of an OLED display panel.

[0096] In the embodiment of the present application, for Mura defects in the production process of an OLED display panel, first, relevant defect records are called from an online monitoring system, and defect detection results are obtained according to the defect records, identification information of a defect panel is sorted out, and its associated processing batch and production history data are associated; then, image information and detection time stamp in the defect detection results are combined to further confirm a specific process station where the abnormal product is identified in the production process and a time when the defect is generated; finally, by comparing production input information of the defect panel and all products in the same batch, total input quantity of the batch of products and the number of defect panels are counted. Meanwhile, according to manual annotation results, the type (such as point, stripe, block, etc.) and spatial position coordinates of each Mura defect are extracted, please refer to Table 1, which is an example of sorting of relevant defect data.

[0097]

[0098] Table 1

[0099] In the embodiment of the present application, the original image can be automatically collected at the same time interval, wherein the original image is saved in BMP format, and the file name of the collected original image is L3LC44E4J151AN0_35000_SVI_07_L32_CAM00.BPM, and relevant defect data (product ID, shooting time, equipment number, etc.) are attached.

[0100] S20: A double-channel preprocessing strategy is adopted.

[0101] S20A: Histogram image conversion.

[0102] ① Color to gray: using weighted average algorithm r j = 0.299 x R + 0.587 x G + 0.114 x B, the original image is color image gray, and then a gray image is obtained. Specifically, the conversion example of part of the pixels in the image L3LC44E4J151AN0_35000_SVI_07_L32_CAM00.BMP is shown in Table 2. Table 2 is a pixel conversion example.

[0103] Pixel position R G B Calculated gray value (1,1) 120 130 110 ≈124 (1,2) 140 135 120 ≈135 (2,1) 100 115 105 ≈110

[0104] Table 2

[0105] ② Adaptive histogram equalization: the gray image is divided into 8 x 8 pixel blocks, and the probability distribution of each small block gray value is calculated according to the calculation formula of the probability distribution of gray value, and then the probability distribution of gray value is mapped to the range of 0-255, and after the fusion processing of each small block, the histogram enhancement image: L3LC44E4J151AN0_35000_SVI_07_L32_CAM00_Hist.PNG is obtained; Finally, the histogram enhancement image is stored.

[0106] S20B: Pseudo-infrared image conversion.

[0107] ① Contrast and brightness adjustment: using contrast and brightness adjustment algorithm to adjust the contrast and brightness of the original image obtained;

[0108] For example, using the formula

[0109] ​②Pseudo-thermal color mapping: using the applyColorMap function of OpenCV, L3LC44E4J151AN0_35000_SVI_07_L32_CAM00.BMP is mapped according to the JET color map. The mapping rule is: low gray corresponds to blue / purple tone; high gray corresponds to red / yellow tone; after mapping, the pseudo-thermal image L3LC44E4J151AN0_35000_SVI_07_L32_CAM00_Thermal.PNG is obtained, which can clearly see that the defect area presents obvious warm color signal; finally, the pseudo-thermal image is saved, and all processing parameters are stored in the database to ensure data traceability. Please refer to Table 3 for pseudo-thermal conversion example data.

[0110]

[0111] Table 3

[0112] S30: Three-modal feature fusion.

[0113] Specifically, the pre-processed image and the original image are fused using a deep network to provide rich information for defect segmentation. The pre-processed image and the original image include:

[0114] L3LC44E4J151AN0_35000_SVI_07_L32_CAM00.BMP

[0115] L3LC44E4J151AN0_35000_SVI_07_L32_CAM00_Hist.PNG

[0116] L3LC44E4J151AN0_35000_SVI_07_L32_CAM00_Thermal.BMP

[0117] Please refer to Figure 2 , Figure 2 is a schematic diagram of three-modal image contrast, wherein (a) is a schematic diagram of the original image, (b) is a histogram-enhanced image, and (c) is a pseudo-thermal image.

[0118] ①Size normalization is performed on the three images to ensure that the spatial resolution is 224x224 pixels; please refer to Table 4 for three modal examples.

[0119] Modality Original resolution Normalized resolution Channel number R (original image) 1280×720 224×224 3 D (histogram enhanced image) 1280×720 224×224 3 T (pseudo-infrared image) 1280×720 224×224 3

[0120] Table 4

[0121] ②The three images are sequentially spliced in the channel dimension; please refer to Table 5 for channel splicing examples.

[0122]

[0123] Table 5

[0124] ③Using 3x3 convolution to compress the channel number to 256, and then the defect integrated image can be obtained to match the subsequent network structure requirements. Please refer to Table 6, which is a channel splicing example. Please refer to Table 7, which is a subsequent network structure requirement.

[0125]

[0126] Table 6

[0127] Feature layer Size Channel number Average activation value T1 64×64 32 0.12 T2 32×32 64 0.08 T3 16×16 128 0.05 T4 8×8 256 0.03 T5 4×4 512 0.02

[0128] Table 7

[0129] In the embodiment of the present application, Mura defects can be obtained through the subsequent network structure. Please refer to

[0130] Table 8, which is a Mura defect example.

[0131]

[0132] Table 8

[0133] And by defect detection on the original image and the defect integrated image respectively, the defect detection effect can be compared. Please refer to Table 9, which is a defect detection effect comparison example.

[0134] Indicator Detection result of original image Detection result of image of the present application Defect accuracy (example) 73.2％ 91.5％ Defect omission rate (example) 35.5％ 8.5％ Defect edge definition (example) Weak Extremely high

[0135] Table 9

[0136] In summary, the present application is based on the multi-modal fusion preprocessing method of pseudo thermal image, histogram image and original defect image fusion. Without additional hardware investment and without increasing the system complexity, more discriminative sample data can be generated, which helps the subsequent defect detection algorithm to more accurately locate and identify Mura defects, thereby improving the product yield.

[0137] In addition, in one embodiment, based on the same inventive idea as the foregoing embodiment, the embodiment of the present application provides an OLED display panel Mura defect image enhancement system. The system corresponds to the method of embodiment 1 one by one. The system comprises:

[0138] An original image acquisition unit, the original image acquisition unit is used to acquire the original image of the OLED display panel, wherein the original image has Mura defects;

[0139] The dual-channel preprocessing unit adopts a dual-channel preprocessing strategy to perform histogram image conversion and pseudo-infrared image conversion on the original image to obtain two complementary histogram enhanced images and a pseudo-infrared image; wherein the histogram image conversion process includes color image graying, local histogram statistics, equalization mapping and smoothing fusion, and the pseudo-infrared image conversion process includes contrast and brightness adjustment, pseudo-infrared color mapping;

[0140] The three-modal feature fusion unit is configured to perform three-modal feature fusion on the original image, the histogram enhanced image and the pseudo-infrared image to obtain a defect integrated image.

[0141] It should be noted that the units in the Mura defect image enhancement system in the embodiment correspond one by one to the steps in the Mura defect image enhancement method in the foregoing embodiment, and therefore the specific implementation and the technical effects achieved by the embodiment can refer to the implementation of the Mura defect image enhancement method, which will not be described herein again.

[0142] In addition, in an embodiment, the present application further provides a computer device, which comprises a processor, a memory and a computer program stored in the memory, and the computer program realizes the method in the foregoing embodiment when executed by the processor.

[0143] In addition, in an embodiment, the present application further provides a computer storage medium, which stores a computer program, and the computer program realizes the method in the foregoing embodiment when executed by a processor.

[0144] In some embodiments, the computer readable storage medium can be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface memory, optical disc or CD-ROM, etc.; or can be various devices including one or any combination of the above memories. The computer can be various computing devices including smart terminals and servers.

[0145] In some embodiments, the executable instructions can be in the form of programs, software, software modules, scripts or codes, written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and can be deployed in any form, including being deployed as independent programs or as modules, components, subroutines or other units suitable for use in a computing environment.

[0146] As an example, executable instructions can correspond to a file in a file system, but in many cases will reside in a portion of the main memory (e.g., random access memory) during execution; in some cases, the executable instructions can also reside on secondary storage (e.g., a disk) from which they are loaded into memory at runtime. As an example, executable instructions can be stored in a compressed, unidentified form and then decompressed and / or identified (e.g., decrypted) prior to execution.

[0147] As an example, executable instructions can be deployed on one computing device, or on multiple computing devices located at one site, or distributed across multiple sites and interconnected by a communication network.

[0148] It has to be noted that, as used herein, the terms "includes" and / or "contains", or any other tautological variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements is not required to comprise only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a" does not, without further constraints, exclude the presence of additional identical elements in the process, method, article, or apparatus that comprises the element.

[0149] The above-mentioned sequence numbers of the embodiments of the present application are only for description, and do not represent advantages or disadvantages of the embodiments.

[0150] Those skilled in the art can clearly understand the above-mentioned embodiment method by means of software and necessary general hardware platform, of course, it can also be realized by hardware, but in many cases, the former is a better embodiment. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a storage medium (such as a read-only memory / random access memory, a magnetic disk, an optical disk), and includes a plurality of instructions for causing a multimedia terminal device (which can be a mobile phone, a computer, a television receiver, or a network device, etc.) to execute the method described in each embodiment of the present application.

[0151] The above is only the preferred embodiment of the present application, and does not limit the patent scope of the present application, and any equivalent structure or equivalent process transformation using the content of the specification and drawings of the present application, or direct or indirect application in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for enhancing images of Mura defects in an OLED display panel, characterized in that, The method includes the following steps: Obtain an original image of an OLED display panel, wherein the original image contains a Mura defect; A dual-channel preprocessing strategy is used to perform histogram image conversion and pseudo-infrared image conversion on the original image to obtain two complementary histogram-enhanced images and pseudo-infrared images. The histogram image conversion process includes color image grayscale conversion, local histogram statistics, equalization mapping, and smooth fusion. The pseudo-infrared image conversion process includes contrast and brightness adjustment and pseudo-infrared color mapping. Three-modal feature fusion is performed on the original image, histogram-enhanced image, and pseudo-infrared image to obtain a defect-integrated image.

2. The method for enhancing Mura defect images in an OLED display panel according to claim 1, characterized in that, The histogram image conversion process is as follows: A weighted average algorithm is used to convert the original image to grayscale to obtain a single-channel grayscale image; A single-channel grayscale image is divided into several small blocks, and the probability distribution of grayscale values ​​is calculated within each small block to achieve local histogram statistics. Based on the statistical results of the local histogram, the gray value mapping of each small block is extended to the full gray level range to achieve local equalization mapping; Based on the local equalization mapping results, several small blocks are smoothly fused to generate an overall histogram-enhanced image.

3. The method for enhancing Mura defect images in an OLED display panel according to claim 1, characterized in that, The process of pseudo-infrared image conversion is as follows: The contrast and brightness of the original image are adjusted using contrast and brightness adjustment algorithms to obtain an adjusted image with local grayscale changes; JET color mapping is performed on the adjusted image with local grayscale changes to map the grayscale values ​​to a gradient color space to obtain a pseudo-infrared image.

4. The method for enhancing Mura defect images in an OLED display panel according to claim 1, characterized in that, The process of the three-modal feature fusion is as follows: The original image, histogram-enhanced image, and pseudo-infrared image are size-normalized to ensure that the spatial resolution of the images is the same. Channel stitching is performed on the original image, histogram-enhanced image, and pseudo-infrared image with the same resolution along the channel dimension to obtain a multi-channel fused feature image; Dimensionality reduction is performed on the multi-channel fused feature image to obtain the defect integrated image.

5. The method for enhancing Mura defect images in an OLED display panel according to claim 1, characterized in that, The formula for the weighted average algorithm is as follows: r j =0.299×R+0.587×G+0.114×B Where, r j This represents the grayscale value of a pixel location in the original image, while RGB represents the three primary colors of a pixel location in the original image.

6. The method for enhancing Mura defect images in an OLED display panel according to claim 5, characterized in that, The formula for calculating the probability distribution of gray values ​​within the small block is: Where p(r) j ) represents the probability distribution of grayscale values ​​within each small block, n j This indicates that the grayscale value within the small block is r. j The number of pixels, where n represents the total number of pixels within the block.

7. The method for enhancing Mura defect images in an OLED display panel according to claim 6, characterized in that, The calculation formula for the smooth fusion of the small blocks is as follows: in, This represents a histogram-enhanced image, where L represents the total number of gray levels, r represents the gray value of a pixel in the original image, and p(r) j ) represents the probability distribution of grayscale values ​​within each small block.

8. The method for enhancing Mura defect images in an OLED display panel according to claim 1, characterized in that, The calculation formula for the contrast and brightness adjustment algorithm is as follows: in, This indicates adjusting the image, where I(x,y) represents the grayscale intensity at the pixel position (x,y). β represents the contrast adjustment factor, β represents the brightness compensation parameter, and the CLIP(·) function ensures that the pixel values ​​for contrast and brightness adjustment remain within the effective range.

9. A system for enhancing images of Mura defects in an OLED display panel, characterized in that, The system includes: A raw image acquisition unit is used to acquire a raw image of an OLED display panel, wherein the raw image contains a Mura defect; A dual-channel preprocessing unit employs a dual-channel preprocessing strategy to perform histogram image conversion and pseudo-infrared image conversion on the original image, thereby obtaining two complementary histogram-enhanced images and pseudo-infrared images. The histogram image conversion process includes color image grayscale conversion, local histogram statistics, equalization mapping, and smooth fusion. The pseudo-infrared image conversion process includes contrast and brightness adjustment and pseudo-infrared color mapping. The three-modal feature fusion unit is used to perform three-modal feature fusion on the original image, histogram enhanced image and pseudo-infrared image to obtain a defect integrated image.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes a computer program, it implements the method for enhancing Mura defect images in an OLED display panel as described in any one of claims 1-8.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method for enhancing Mura defect images in an OLED display panel according to any one of claims 1-8.

Images