Mura defect detection method and system based on multi-domain fusion

Abstract

The invention provides a Mura defect detection method and system based on multi-domain fusion, and relates to the technical field of defect detection, and the method comprises the steps: carrying out the preprocessing of an RCB image of a display panel, so as to obtain an RGB standard image; performing HSV space conversion on the RGB standard image to obtain an H-channel image, an S-channel image and a V-channel image; performing domain optimization and multi-domain fusion on the H-channel image, the S-channel image and the V-channel image to obtain a defect difference image; and performing adaptive threshold segmentation on the defect difference image to obtain a Mura defect positioning result. In order to enhance Mura defect contrast and prevent imaging interference of a display panel, firstly, channel splitting is carried out on an RGB standard image, then domain division optimization and multi-domain fusion are carried out on three channel images, and finally, adaptive threshold segmentation is carried out on the multi-domain fused image, so that the Mura defect position can be highlighted; the problems that existing Mura defect detection is low in efficiency and high in omission ratio and over-detection rate are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of defect detection, in particular to a Mura defect detection method and system based on multi-domain fusion. BACKGROUND

[0002] Display panels have been widely used in high-end consumer electronics due to their excellent display performance, but display panels often have Mura defects. Mura defects are a unique flaw in the manufacturing of display devices (such as LCD and OLED screens), which are characterized by uneven brightness or chroma in local areas of the screen, forming visible patches, clouds or streaks. It usually contains the following characteristics: 1. Low contrast, i.e. the brightness difference between the defect and the background is usually <10% (the human eye can detect a threshold of about 5%); 2. Blurred boundaries, i.e. the transition between the defect area and the normal area is gradual, with no clear edges; 3. Low spatial frequency domain, i.e. it appears as a large area of slowly changing texture (as opposed to high-frequency pixel defects); 4. Strong viewing angle dependence, some Mura defects are only visible at a specific viewing angle. These Mura defects are usually detected by the naked eye during the production process. However, the manual detection method is relatively costly and is subject to human subjective factors, which may result in over-detection or missed detection.

[0003] In order to save manpower and improve the detection efficiency of Mura defects, a traditional visual detection method is usually used to detect Mura defects. However, the traditional visual detection method is sensitive to light and cannot be used on machines with variable light sources. SUMMARY

[0004] In order to solve the problems of low detection efficiency, high over-detection rate and high missed detection rate of existing Mura defect detection, the present application provides a Mura defect detection method and system based on multi-domain fusion.

[0005] In a first aspect, the present application provides a Mura defect detection method based on multi-domain fusion, which comprises the following steps:

[0006] An RCB image of the display panel is collected, and the RCB image of the display panel is preprocessed to obtain an RGB standard image;

[0007] The RGB standard image is converted to HSV space and separated into H, S and V channels to obtain H, S and V channel images;

[0008] The H, S and V channel images are optimized and fused in multiple domains to obtain a defect difference image;

[0009] The defect feature enhancement, display feature optimization and adaptive threshold segmentation are performed on the defect difference image to obtain a Mura defect positioning result.

[0010] In the above embodiment, in order to effectively extract the features between channels, enhance the contrast of Mura defects and prevent imaging interference of the display panel, the three channels of the HSV image are split, and the three channels are respectively optimized, and the feature maps of the three channels are fused and differenced, which is beneficial to highlight the Mura defect position.

[0011] As some optional embodiments of the present application, the process of pre-processing the RCB image of the display panel is as follows:

[0012] The display area is located by using Canny edge detection or template matching, and the non-display area interference is removed according to the positioning result;

[0013] The RCB image after removing the non-display area interference is corner positioned and rotation corrected to retain the RCB image in the horizontal or vertical direction;

[0014] The rotation corrected RGB image is white balance corrected, and the image is resolution normalized by using bilinear interpolation to scale the image to a standard size to obtain an RGB standard image.

[0015] In the above embodiment, the RCB image of the display panel is pre-processed, which can provide a more standard image for subsequent Mura defect detection.

[0016] As some optional embodiments of the present application, the process of performing HSV space conversion and HSV space channel separation on the RGB standard image is as follows:

[0017] The RGB standard image is converted into an HSV standard image by using a space conversion formula;

[0018] The HSV standard image is channel separated, and each channel is saved as a separate image to obtain an H channel image, an S channel image and a V channel image.

[0019] In the above embodiment, the HSV space conversion involved in the present application is to decompose the color into hue, saturation and lightness three independent dimensions by simulating the human visual system, and each channel can be optimized for image processing effect after channel separation.

[0020] As some optional embodiments of the present application, the process of performing domain optimization and multi-domain fusion on the H channel image, the S channel image and the V channel image is as follows:

[0021] Discrete cosine transform is performed on the H channel image, low-frequency filtering is adopted to extract low-frequency components from the discrete cosine transformed feature map, and inverse discrete cosine transform is performed on the low-frequency component extracted feature map to obtain an H domain component feature map;

[0022] Gaussian filtering is adopted to perform weak flattening on a high-texture region and strong flattening on a flat region of the S channel image, singular value decomposition is performed on the feature map of the weak flattening on the high-texture region and the strong flattening on the flat region, and low-rank background reconstruction is performed on the singular value decomposed feature map to obtain an S domain component feature map;

[0023] Fast Fourier transform is performed on the V channel image, high-frequency filtering is adopted to extract high-frequency components from the fast Fourier transformed feature map, frequency domain band-pass filtering is performed on the high-frequency component extracted feature map, and inverse fast Fourier transform is performed on the frequency domain band-pass filtered feature map to obtain a V domain component feature map;

[0024] Feature fusion is performed on the H domain component feature map and the S domain component feature map to obtain an HS domain fused image, and difference processing is performed on the HS domain fused image and the V domain component feature map to obtain a defect difference image.

[0025] In the above embodiments, the present application can correct light source uniformity, suppress chroma change, prevent light interference, and eliminate the gradual background caused by non-uniform backlight during the image acquisition process of the display panel by performing domain optimization on the H channel image; the present application can correct visual effect, separate background texture, eliminate the influence of structural noise, and solve the problem of low saturation of Mura defects by performing domain optimization on the S channel image; and the present application can suppress reflection, eliminate the gradual background caused by reflection during the image acquisition process of the display panel, enhance low contrast, convert the luminance-chroma separation perception mechanism of the human eye into a calculable frequency domain processing chain, realize the alignment of machine vision and human eye judgment, and highlight Mura defects by performing domain optimization on the V channel image.

[0026] As some optional embodiments of the present application, the process of defect feature enhancement on the defect difference image is as follows:

[0027] The defect difference image is divided into N×N sub-blocks, and histogram equalization is performed on each sub-block to limit the contrast improvement amplitude;

[0028] Local standard deviation calculation is performed on the histogram equalized defect difference image, and local gain is performed according to the local standard deviation calculation result to highlight potential defects.

[0029] In the above embodiments, the present application performs local gain according to the local standard deviation calculation result, quantifies local gray scale change, highlights defect edges and texture details, and highlights potential Mura defects.

[0030] As some optional embodiments of the present application, the process of display feature optimization of the defect difference image is as follows:

[0031] The open operation and the close operation are performed on the defect difference image to remove small area defects and fill defect holes;

[0032] Gaussian edge blurring is performed on the defect difference image after the open operation and the close operation to avoid color step fault and transition sharpening.

[0033] In the above embodiment, the open operation is performed by first etching and then expanding to remove small noise points in the image while maintaining the shape of larger defects; the close operation is performed by first expanding and then etching to fill the holes inside the defects or connect the disconnected areas to form a more complete area.

[0034] As some optional embodiments of the present application, the process of adaptive threshold segmentation of the defect difference image is as follows:

[0035] The local mean and the local standard deviation of the defect difference image are obtained, and an adaptive threshold parameter is constructed according to the local mean and the local standard deviation of the defect difference image;

[0036] The adaptive threshold segmentation is performed on the defect difference image according to the adaptive threshold parameter to obtain the Mura defect positioning result.

[0037] In the above embodiment, the adaptive threshold segmentation is used for defect detection, which has high detection efficiency and low false detection rate and over-detection rate.

[0038] In a second aspect, the present application provides a Mura defect detection system based on multi-domain fusion, which comprises:

[0039] An image preprocessing unit, which is used for acquiring an RCB image of a display panel and pre-processing the RCB image of the display panel to obtain an RGB standard image;

[0040] An HSV space conversion unit, which is used for performing HSV space conversion and HSV space channel separation on the RGB standard image to obtain an H channel image, an S channel image and a V channel image;

[0041] A multi-domain fusion unit, which is used for performing domain optimization and multi-domain fusion on the H channel image, the S channel image and the V channel image to obtain a defect difference image;

[0042] A Mura defect positioning unit, which is used for performing defect feature enhancement, display feature optimization and adaptive threshold segmentation on the defect difference image to obtain a Mura defect positioning result.

[0043] 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 executes the Mura defect detection method based on multi-domain fusion.

[0044] 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 detection method based on multi-domain fusion.

[0045] The beneficial effects of the present application are as follows: according to the imaging characteristics of the display panel and the defect characteristics of the Mura defect, in order to effectively extract the characteristics between channels, enhance the contrast of the Mura defect, and prevent the imaging interference of the display panel, the image is split by channels, and is respectively optimized, and the feature maps of three channels are fused and differentiated, which is conducive to highlighting the Mura defect position, and the detection efficiency is high, and the false negative rate and the false positive rate are low. BRIEF DESCRIPTION OF DRAWINGS

[0046] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used 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 to the scope. For those skilled in the art, other related drawings can also be obtained without creative labor.

[0047] Figure 1 is a flow chart of the Mura defect detection method based on multi-domain fusion according to some embodiments of the present application. DETAILED DESCRIPTION

[0048] In order to better understand the above technical solutions, the technical solutions of the present application will be described in detail below through the drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical solutions of the present application, and are not limitations to the technical solutions of the present application. In the case of no conflict, the technical features in the embodiments of the present application and the embodiments can be combined with each other.

[0049] It should also be understood that, in order to simplify the expression disclosed by the present application, and to help the understanding of at least one embodiment of the present application, in the foregoing description of the embodiments of the present application, sometimes multiple features are combined into one embodiment, drawing or description thereof. However, this disclosure method does not mean that the features required by the object of the present application are more than the features mentioned in the claims. In fact, the features of the embodiments are less than all the features of the disclosed single embodiment.

[0050] Example 1

[0051] This invention provides a Mura defect detection method based on multi-domain fusion. Please refer to [link / reference]. Figure 1 , Figure 1 This is a flowchart of the Mura defect detection method based on multi-domain fusion, the method comprising the following steps:

[0052] (1) Use a high-resolution industrial camera (such as a line scan camera CCD or an area scan camera CMOS) or a professional scanner to acquire the RCB image of the display panel under uniform lighting conditions, and preprocess the RCB image of the display panel to obtain the RGB standard image.

[0053] In this embodiment of the invention, the process for preprocessing the RCB image of the display panel is as follows:

[0054] (1.1) The display area is located using Canny edge detection or template matching, and non-display area interference is eliminated based on the location results. Among them, Canny edge detection or template matching can achieve accurate edge location, thereby locating the display area in the middle of the image, and then eliminating non-realistic areas outside the display area, providing a more standardized image for subsequent Mura defect detection.

[0055] (1.2) Corner localization and rotation correction are performed on the RCB images that have been removed from non-display area interference to preserve the RCB images in the horizontal or vertical direction. Among them, rotation correction uses affine transformation or rotation matrix to rotate the image to the correct angle to adjust the tilted image to the standard orientation (e.g., horizontal / vertical) and preserve the RCB images in the horizontal or vertical direction, so as to provide more standard images for subsequent Mura defect detection.

[0056] (1.3) White balance correction is performed on the rotation-corrected RGB image, and bilinear interpolation is used to normalize the resolution of the white balance-corrected image, scaling the image to a standard size to obtain a standard RGB image. White balance correction effectively eliminates color cast, improving the visual effect of the image and the accuracy of subsequent processing. Simultaneously, resolution normalization scales the image to a standard size, effectively eliminating size differences and improving the robustness and accuracy of subsequent algorithms.

[0057] In this embodiment of the invention, the imaging characteristics of the display panel mainly include uneven light source distribution, viewing angle effect, and surface reflection; while the defect characteristics of Mura defects mainly include low contrast, blurred boundaries, and diverse shapes. Combining the imaging characteristics of the display panel and the defect characteristics of Mura defects, the RGB standard image is first converted to HSV space. Then, domain-specific optimization and multi-domain fusion are performed on the imaging characteristics of the display panel and the defect characteristics of Mura defects to achieve light source uniformity correction, viewing angle effect correction, reflection suppression, and Mura defect enhancement, thereby achieving effective detection of Mura defects.

[0058] (2) Perform HSV space conversion and HSV space channel separation on the RGB standard image to obtain H channel image, S channel image and V channel image. HSV space conversion simulates the human visual system to decompose color into three independent dimensions: hue, saturation and lightness. After channel separation, the image processing effect of each channel can be optimized in a targeted manner.

[0059] In this embodiment of the invention, the process of performing HSV spatial conversion and HSV spatial channel separation on an RGB standard image is as follows:

[0060] (2.1) The RGB standard image is converted to HSV space using a spatial transformation formula to obtain the HSV standard image. The HSV spatial transformation of the RGB standard image can achieve accurate conversion from RGB to HSV, providing an ideal image input for subsequent Mura defect detection.

[0061] (2.2) Channel separation is performed on the HSV standard image, saving each channel as a separate image to obtain the H-channel image, S-channel image, and V-channel image. Specifically, the H-channel (hue), S-channel (saturation), and V-channel (lightness) are three independent dimensions of grayscale image representation in the HSV color space. They describe color information from different perspectives and have fundamental differences in visual representation and practical applications. Therefore, subsequent optimization and fusion of the H-channel image, S-channel image, and V-channel image separately can achieve light source uniformity correction, viewing angle effect correction, reflection suppression, and Mura defect enhancement.

[0062] Specifically, the space transformation formula is as follows:

[0063] V = max(R, G, B)

[0064]

[0065] (3) Perform domain-specific optimization and multi-domain fusion on the H-channel image, S-channel image and V-channel image to obtain the defect difference image.

[0066] In this embodiment of the invention, the process of performing domain-specific optimization and multi-domain fusion on the H-channel image, S-channel image, and V-channel image is as follows:

[0067] (3.1) In order to achieve uniformity correction of the light source and eliminate the gradient background caused by the uneven backlight during the image acquisition process of the display panel, the Mura defect is highlighted. The Mura defect is characterized by low-frequency perturbation in the color space. In order to effectively separate the stable low-frequency features and high-frequency perturbation random noise in the color channel, it is necessary to perform discrete cosine transform on the H channel image, and then use Butterworth low-frequency filtering to extract the low-frequency components of the feature map of the discrete cosine transform. The key reason for using discrete cosine transform is that its decomposition characteristics are more suitable for processing the gradually changing Mura defect features. In order to eliminate the gradient background caused by uneven background light source, the low-frequency background components are effectively extracted and the feature map is smoothly transitioned. Finally, the feature map of the low-frequency component extraction is subjected to inverse discrete cosine transform to obtain the H domain component feature map, ensuring that the feature map of the inverse transform only contains low-frequency components, which is convenient for subsequent image fusion and image difference processing.

[0068] Specifically, for an image patch of size m×n, the formula for the two-dimensional discrete cosine transform is as follows:

[0069]

[0070] Among them, H det (u,v) represents the discrete cosine transform coefficients, and H(x,y) represents the pixels of the H-channel image.

[0071] (3.2) To achieve visual effect correction and address the issue of low saturation in Mura defects, Gaussian filtering was first applied to the S-channel image to achieve weak smoothing in high-texture regions and strong smoothing in flat regions. Singular value decomposition was then performed on the feature maps of these regions, followed by low-rank background reconstruction. Hierarchical enhancement was then applied to low-saturation regions while maintaining color continuity to obtain S-domain component feature maps, which facilitates local image processing during subsequent defect feature enhancement.

[0072] (3.3) To achieve reflection suppression and eliminate the gradient background caused by reflection during the image acquisition process of the display panel, a Fast Fourier Transform (FFT) is performed on the V-channel image. High-frequency filtering is used to extract high-frequency components from the feature map of the FFT, and frequency domain bandpass filtering is performed on the feature map of the high-frequency components. Through frequency domain operations, the extraction and preservation of detail information are precisely controlled to balance the needs of detail enhancement and noise suppression. Furthermore, an inverse FFT is performed on the feature map of the frequency domain bandpass filter to obtain the V-domain component feature map.

[0073] In this embodiment of the invention, before performing a fast Fourier transform on the V-channel image, it is necessary to enhance the feature defects of the V-channel image, that is, to use a Hanning window to reduce spectral leakage and aliasing effects.

[0074] Specifically, the formula for the Hanning window is as follows:

[0075]

[0076] Where w(x,y) represents the temporal representation of the Hanning window for a pixel, and m×n represents the size of the image patch, which also represents the size of the Hanning window.

[0077] (3.4) Feature fusion is performed on the H-domain and S-domain component feature maps to obtain an HS-domain fused image. Feature fusion of the H-domain and S-domain can significantly improve the performance of image processing tasks through the complementarity of color type and purity. Specifically, channel splicing or element-wise addition can be used to achieve feature fusion, and the fused feature map can capture both color essence and purity information simultaneously. Further, differential processing is performed on the HS-domain fused image and the V-domain component feature map to obtain a defect differential image. Differential processing of the HS-domain fused image and the V-domain component feature map can achieve detail enhancement and anomaly detection through the comparison of color-brightness information differences. Specifically, pixel-wise differential or gradient differential can be used, and the differential feature map can simultaneously highlight color-saturation features and brightness features.

[0078] (4) Perform defect feature enhancement, display feature optimization and adaptive threshold segmentation on the defect difference image to obtain the Mura defect localization result.

[0079] In this embodiment of the invention, the process of enhancing defect features in a defect difference image is as follows:

[0080] (4.1) Divide the defect difference image into N×N sub-blocks, perform histogram equalization on each sub-block to enhance the contrast between bright and dark defects, and limit the contrast enhancement by limiting the histogram cropping threshold.

[0081] (4.2) Calculate the local standard deviation of the histogram equalized defect difference image, and perform local gain based on the local standard deviation calculation results to quantify the local grayscale changes, highlight the defect edges and texture details, and highlight the potential Mura defects.

[0082] Specifically, the process for optimizing the display features of the defect difference image is as follows:

[0083] (4.3) Opening and closing operations are performed on the defect difference image to remove small-area defects and fill defect voids. Specifically, the opening operation removes small noise points in the image by first eroding and then dilating, while maintaining the shape of larger defects; the closing operation fills voids or disconnected areas inside the defect by first dilating and then eroding, forming a more complete area.

[0084] (4.4) Gaussian edge blur is applied to the defect difference image of opening and closing operations. Gaussian blur makes the edges smoother and reduces abrupt changes to avoid color level breaks and over-sharpening.

[0085] Specifically, the process of adaptive thresholding segmentation for defect difference images is as follows:

[0086] (4.5) Obtain the local mean and local standard deviation of the defect difference image, and construct an adaptive threshold parameter based on the local mean and local standard deviation of the defect difference image.

[0087] (4.6) Perform adaptive threshold segmentation on the defect difference image according to the adaptive threshold parameter to obtain the Mura defect localization result.

[0088] Specifically, the formula for constructing the adaptive threshold parameter is as follows:

[0089]

[0090] Where μ(x,y) represents the local mean, σ(x,y) represents the local standard deviation, R represents the dynamic threshold (e.g., 128), and k represents the adjustment parameter (typically 0.2 to 0.5).

[0091] In summary, based on the imaging characteristics of the display panel and the defect characteristics of the Mura defect, this invention effectively extracts features between channels, enhances the contrast of the Mura defect, and prevents interference from the display panel's imaging. The image is split into channels and optimized separately. The feature maps of the three channels are fused and differentiated, which helps to highlight the location of the Mura defect, resulting in high detection efficiency and low false negative and false positive rates.

[0092] Example 2

[0093] This invention provides a Mura defect detection system based on multi-domain fusion, the system comprising:

[0094] An image preprocessing unit is used to acquire the RCB image of the display panel and preprocess the RCB image of the display panel to obtain an RGB standard image.

[0095] The HSV spatial conversion unit is used to perform HSV spatial conversion and HSV spatial channel separation on RGB standard images to obtain H-channel images, S-channel images and V-channel images.

[0096] A multi-domain fusion unit is used to perform domain-specific optimization and multi-domain fusion on the H-channel image, S-channel image, and V-channel image to obtain a defect difference image.

[0097] The Mura defect localization unit is used to perform defect feature enhancement, display feature optimization, and adaptive threshold segmentation on the defect difference image to obtain the Mura defect localization result.

[0098] Example 3

[0099] This invention provides a computer device including a memory and a processor. The memory stores a computer program, which executes the Mura defect detection method based on multi-domain fusion as described in Embodiment 1 when the processor is running. The computer device provided in this embodiment can implement the method described in Embodiment 1; to avoid repetition, it will not be described again here.

[0100] Example 4

[0101] This invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the Mura defect detection method based on multi-domain fusion as described in Embodiment 1. The computer-readable storage medium provided in this embodiment can implement the method described in Embodiment 1; to avoid repetition, it will not be described again here.

[0102] The processor can be a central processing unit (CPU), or other general-purpose processors, digital signal processors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0103] The memory can be used to store the computer program and / or modules. The processor implements various functions of the Mura defect detection system based on multi-domain fusion in the invention by running or executing the data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, RAM, plug-in hard disk, smart memory card, secure digital card, flash memory card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0104] If the Mura defect detection system based on multi-domain fusion is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program that can be stored in a computer-readable storage medium. When executed by a processor, this computer program can implement the steps of the various method embodiments described above. The computer program includes computer program code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory, random access memory, dot carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction.

[0105] The basic concepts of this invention have been described. It is obvious to those skilled in the art that the detailed disclosure above is merely illustrative and does not constitute a limitation of this specification. Although not explicitly stated herein, those skilled in the art may make various modifications, improvements, and corrections to this specification. Such modifications, improvements, and corrections are suggested in this specification and therefore remain within the spirit and scope of the exemplary embodiments described herein.

[0106] Furthermore, this specification uses specific terms to describe embodiments thereof. For example, "an embodiment," "one embodiment," and / or "some embodiments" refer to a particular feature, structure, or characteristic associated with at least one embodiment of this specification. Therefore, it should be emphasized and noted that references to "an embodiment," "one embodiment," or "an alternative embodiment" in different locations throughout this specification do not necessarily refer to the same embodiment. Moreover, certain features, structures, or characteristics in one or more embodiments of this specification can be appropriately combined.

[0107] Furthermore, those skilled in the art will understand that various aspects of this specification can be described and illustrated in several patentable ways or situations, including any new and useful combination of processes, machines, products, or substances, or any new and useful improvements thereof. Accordingly, various aspects of this specification can be implemented entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. All of the above hardware or software may be referred to as a “data block,” “module,” “engine,” “unit,” “component,” or “system.” Furthermore, various aspects of this specification may be represented as a computer product located on one or more computer-readable media, including computer-readable program code.

[0108] Computer storage media may contain a propagated data signal containing computer program code, for example, on baseband or as part of a carrier wave. This propagated signal may take various forms, including electromagnetic, optical, and suitable combinations thereof. Computer storage media can be any computer-readable medium other than a computer-readable storage medium, which can be connected to an instruction execution system, apparatus, or device to enable communication, propagation, or transmission of a program for use. The program code located on the computer storage medium can be propagated through any suitable medium, including radio, cable, fiber optic cable, RF, or similar media, or any combination of the above media.

Claims

1. A Mura defect detection method based on multi-domain fusion, characterized in that, The method includes the following steps: The RCB image of the display panel is acquired and preprocessed to obtain the RGB standard image. Perform HSV spatial transformation and HSV spatial channel separation on RGB standard images to obtain H channel images, S channel images, and V channel images; Domain-specific optimization and multi-domain fusion are performed on the H-channel, S-channel, and V-channel images to obtain defect difference images; Defect feature enhancement, display feature optimization, and adaptive threshold segmentation are performed on the defect difference image to obtain the Mura defect localization result.

2. The Mura defect detection method based on multi-domain fusion according to claim 1, characterized in that, The preprocessing procedure for the RCB image of the display panel is as follows: The display area is located using Canny edge detection or template matching, and non-display area interference is eliminated based on the location results; Corner point localization and rotation correction are performed on the RCB images that have eliminated interference from non-display areas in order to preserve the RCB images in the horizontal or vertical directions; The rotation-corrected RGB image is white-balanced, and the resolution is normalized by bilinear interpolation. The image is then scaled to a standard size to obtain a standard RGB image.

3. The Mura defect detection method based on multi-domain fusion according to claim 1, characterized in that, The process of performing HSV spatial conversion and HSV spatial channel separation on an RGB standard image is as follows: The RGB standard image is converted to HSV space using a spatial transformation formula to obtain the HSV standard image. Channel separation is performed on the HSV standard image, and each channel is saved as a separate image to obtain the H channel image, S channel image, and V channel image.

4. The Mura defect detection method based on multi-domain fusion according to claim 1, characterized in that, The process of performing domain-specific optimization and multi-domain fusion on H-channel, S-channel, and V-channel images is as follows: The H-channel image is subjected to discrete cosine transform (DCT). Low-frequency filtering is used to extract low-frequency components from the feature map of the DCT. The feature map of the low-frequency components is then subjected to inverse DCT to obtain the H-domain component feature map. Gaussian filtering is used to smooth the high-texture region weakly and the flat region strongly. Singular value decomposition is performed on the feature maps of the high-texture region weakly smoothed and the flat region strongly smoothed, and low-rank background reconstruction is performed on the feature maps of the singular value decomposition to obtain the S-domain component feature maps. A fast Fourier transform is performed on the V-channel image. High-frequency components are extracted from the feature map of the fast Fourier transform using high-frequency filtering. The feature map of the high-frequency components is then subjected to frequency domain bandpass filtering. Finally, an inverse fast Fourier transform is performed on the feature map of the frequency domain bandpass filtering to obtain the V-domain component feature map. Feature fusion is performed on the H-domain component feature map and the S-domain component feature map to obtain the HS-domain fused image. Then, the HS-domain fused image and the V-domain component feature map are differentially processed to obtain the defect differential image.

5. The Mura defect detection method based on multi-domain fusion according to claim 1, characterized in that, The process for enhancing defect features in a defect difference image is as follows: The defect difference image is divided into N×N sub-blocks, and histogram equalization is performed on each sub-block to limit the contrast enhancement. Local standard deviation is calculated for the histogram-equalized defect difference image, and local gain is applied based on the local standard deviation calculation results to highlight potential defects.

6. The Mura defect detection method based on multi-domain fusion according to claim 1, characterized in that, The process for optimizing the display features of the defect difference image is as follows: Opening and closing operations are performed on the defect difference image to remove small-area defects and fill defect voids; Gaussian edge blurring is applied to the defective difference image of opening and closing operations to avoid color level breaks and over-sharpening.

7. The Mura defect detection method based on multi-domain fusion according to claim 1, characterized in that, The process of adaptive thresholding segmentation for defect difference images is as follows: Obtain the local mean and local standard deviation of the defect difference image, and construct an adaptive threshold parameter based on the local mean and local standard deviation of the defect difference image; Adaptive threshold segmentation is performed on the defect difference image based on the adaptive threshold parameter to obtain the Mura defect localization result.

8. A Mura defect detection system based on multi-domain fusion, characterized in that, The system includes: An image preprocessing unit is used to acquire the RCB image of the display panel and preprocess the RCB image of the display panel to obtain an RGB standard image. The HSV spatial conversion unit is used to perform HSV spatial conversion and HSV spatial channel separation on RGB standard images to obtain H-channel images, S-channel images and V-channel images. A multi-domain fusion unit is used to perform domain-specific optimization and multi-domain fusion on the H-channel image, S-channel image, and V-channel image to obtain a defect difference image. The Mura defect localization unit is used to perform defect feature enhancement, display feature optimization, and adaptive threshold segmentation on the defect difference image to obtain the Mura defect localization result.

9. 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 the computer program, it implements the Mura defect detection method based on multi-domain fusion as described in any one of claims 1-7.

10. 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 Mura defect detection method based on multi-domain fusion as described in any one of claims 1-7.

Images