Outdoor dot matrix display screen picture flaw detection method based on texture features

Abstract

The invention discloses an outdoor dot matrix display screen picture flaw detection method, device and system based on texture features and a server. According to the method, at least two pure color test pictures are obtained, and defect full-coverage detection is realized by utilizing sensitivity differences of different color channels; besides, multi-scale feature extraction is carried out based on a Gabor filter, so that scale features such as single-point dead pixels, regional dark spots and module-level distortion can be captured, and multiple types of flaws such as brightness deviation and large-area uneven brightness can be covered. According to the method, the detection rate of traditional leak detection scenes such as small dead pixels and color deviation is increased to 95% or above, the environment robustness is high, the false detection rate in the complex illumination and vibration environment is low, and a high-reliability automatic defect detection solution is provided for the field of outdoor advertisements and public display.

Description

Technical Field

[0001] The present application relates to the field of image processing technology, and in particular to a method, device, system and server for detecting image defects on an outdoor dot matrix display screen based on texture features. Background Art

[0002] Dot-matrix displays are widely used in outdoor advertising and public information displays due to their high brightness and wide viewing angles. However, complex outdoor environments (such as drastic temperature fluctuations, dust intrusion, and mechanical vibration) can easily lead to defects such as dead pixels, uneven brightness, and texture distortion. These defects not only affect the visual display but can also accumulate over time, leading to widespread failures, increasing maintenance costs and risking damage to brand image. Therefore, efficient and accurate defect detection on dot-matrix displays is critical to ensuring the proper operation of the equipment.

[0003] Existing remote monitoring methods for defect detection on dot-matrix displays rely on fixed cameras to capture and analyze images. However, large outdoor screens are often installed at long distances, resulting in insufficient image resolution and difficulty identifying small defective pixels. Traditional algorithms rely on a single, solid-color image for detection, which can lead to missed detections due to overlap between the defective pixel's color and the detection color. Unsupervised algorithms based on deep learning achieve defect detection by reconstructing a textured background, but they rely too heavily on reconstruction accuracy, resulting in both missed and overdetected detections and poor robustness. Furthermore, existing solutions generally lack coordinated hardware and algorithm design. For example, fixed camera positions lead to image distortion, and preprocessing steps (such as median filtering and geometric correction) are not fully integrated with feature extraction, making them difficult to adapt to the complex outdoor lighting and diverse screen textures.

[0004] Therefore, how to improve the accuracy of defect detection when dealing with complex scenarios is an issue that technical personnel in this field urgently need to solve. Summary of the Invention

[0005] In view of the above-mentioned defects or deficiencies in the prior art, it is desired to provide a method, device, system and server for detecting defects in an outdoor dot matrix display screen based on texture features, which can improve the accuracy of defect detection when dealing with complex scenes.

[0006] In a first aspect, an embodiment of the present application provides a method for detecting image defects on an outdoor dot matrix display screen based on texture features, comprising: Acquire an original image of a pure color test screen of a dot matrix display screen; wherein the pure color test screen includes at least two pure color screens; Preprocessing the original image to obtain a preprocessed image; Calling a Gabor filter to perform multi-scale texture feature extraction on the pre-processed image to generate multi-scale texture features; Based on the multi-scale texture features, the texture background and the defect foreground are separated by a contrast separation model to generate a separation result; The defective area in the original image is marked based on the separation result.

[0007] In one embodiment, the pure color test image includes a blue image and a white image, and the calling of a Gabor filter to perform multi-scale texture feature extraction on the pre-processed image includes: For the preprocessed white screen image, the first set of Gabor filters is called to extract texture features as the first set of features and generate the original feature map; For the preprocessed blue screen image, calling a second group of Gabor filters to extract texture features as a second group of features; wherein the σ parameter of the second group of Gabor filters is greater than the σ parameter of the first group of Gabor filters; calculating a channel difference response map between the first set of features and the second set of features; The channel difference response map is fused with the original feature map to generate an enhanced multi-scale texture feature as the multi-scale texture feature.

[0008] In one embodiment, the texture feature-based outdoor dot matrix display screen defect detection method further includes: Extracting local texture features of the preprocessed image using local binary patterns; Fusing the local texture feature with the multi-scale texture feature to generate a local fusion feature; Based on the multi-scale texture features, the texture background and the defect foreground are separated by a contrast separation model. Specifically, based on the local fusion features, the texture background and the defect foreground are separated by a contrast separation model.

[0009] In one embodiment, fusing the local texture features with the multi-scale texture features includes: Cascading the local texture features and the multi-scale texture features by channel to generate a multi-dimensional fusion feature; Use attention mechanism to calculate fusion weights; The features in the multi-dimensional fusion feature are weightedly fused based on the fusion weight to serve as the local fusion feature.

[0010] In one embodiment, obtaining an original image of a pure color test screen of a dot matrix display screen includes: The motor drives the telescopic rod to move the camera to a position where it can capture pixel-level details of the dot matrix display screen, and the lens plane is parallel to the screen surface, thereby capturing the original image of the pure color test screen.

[0011] In one embodiment, marking the defective area in the original image based on the separation result includes: performing a morphological opening operation on the separation result to obtain a first processed image; removing falsely detected points whose areas are smaller than a threshold value from the first processed image to obtain a second processed image; Calculating geometric features of defects in the second processed image based on connected component analysis; Classify the defects according to the geometric features and the multi-scale texture features to determine the defect type; Marking boxes of different colors are drawn on the original image according to the defect type.

[0012] In one embodiment, after marking the defective area in the original image based on the separation result, the method further includes: Obtaining defect area data from historical inspections as historical data; Comparing and analyzing the historical data with the current defect area, and calculating the defect diffusion rate; If the defect diffusion speed exceeds the diffusion threshold, a warning message is output.

[0013] In a second aspect, an embodiment of the present application provides a device for detecting defects on an outdoor dot matrix display screen based on texture features, comprising: An image acquisition module is used to acquire an original image of a pure color test screen of a dot matrix display screen; wherein the pure color test screen includes at least two pure color screens; A preprocessing module, configured to preprocess the original image to obtain a preprocessed image; A feature extraction module is used to extract multi-scale texture features from the pre-processed image by calling a Gabor filter to generate multi-scale texture features; A separation processing module, configured to separate the texture background from the defect foreground based on the multi-scale texture features by comparing the separation model and generating a separation result; A defect marking module is used to mark the defect area in the original image based on the separation result.

[0014] In a third aspect, an embodiment of the present application provides a texture feature-based outdoor dot matrix display screen defect detection system, comprising: A movable camera is used to capture an original image of a pure color test screen of a dot matrix display screen; wherein the pure color test screen includes at least two pure color screens; The above-mentioned texture feature-based outdoor dot matrix display screen defect detection device is used for the original image, preprocesses the original image to obtain a preprocessed image, uses a Gabor filter to extract multi-scale texture features from the preprocessed image to generate multi-scale texture features, and based on the multi-scale texture features, separates the texture background from the defect foreground through a comparative separation model to generate a separation result; and marks the defect area in the original image based on the separation result; The user interaction device is used to visually output the original image after marking the defect area.

[0015] In a fourth aspect, an embodiment of the present application provides a server comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the steps of a method for detecting defects in an outdoor dot matrix display screen based on texture features are implemented.

[0016] This application provides a texture-based defect detection method for outdoor dot-matrix displays. This method acquires at least two pure-color test images and leverages the sensitivity differences between different color channels to achieve comprehensive defect detection. Furthermore, multi-scale feature extraction based on Gabor filters captures features ranging from single-point bad pixels, regional dark spots, and module-level distortion, covering a wide range of defect types, from large brightness deviations to large-area uneven brightness. This method increases the detection rate of traditionally missed scenarios, such as small bad pixels and color deviations, to over 95%. Furthermore, it exhibits strong environmental robustness, with a low false positive rate in complex lighting and vibration environments. This method provides a highly reliable automatic defect detection solution for outdoor advertising and public displays.

[0017] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Other features, objects and advantages of the present application will become more apparent upon reading the detailed description of non-limiting embodiments made with reference to the following drawings: Figure 1 A schematic diagram showing a flow chart of a method for detecting defects on an outdoor dot matrix display screen based on texture features provided in an embodiment of the present application is shown; Figure 2 An exemplary structural block diagram of a texture feature-based outdoor dot matrix display screen defect detection device provided in an embodiment of the present application is shown; Figure 3 A schematic diagram of the structure of a computer system suitable for implementing a server according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0019] The present application will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the relevant invention and are not intended to limit the invention. It should also be noted that, for ease of description, only portions relevant to the invention are shown in the accompanying drawings.

[0020] It should be noted that, in the absence of conflict, the embodiments in this application and the features in the embodiments can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments. Although the embodiments of the present application provide the method operation instruction steps shown in the following embodiments or drawings, more or fewer operation instruction steps may be included in the method based on routine or no creative labor. In steps where there is no necessary causal relationship logically, the execution order of these steps is not limited to the execution order provided in the embodiments of the present application. During the actual processing process or when the device is executed, the method may be executed in the order of the methods shown in the embodiments or drawings or in parallel.

[0021] Example 1: This embodiment proposes a method for detecting defects on an outdoor dot matrix display screen based on texture features. Figure 1 , Figure 1 FIG. 1 is a flow chart showing a method for detecting defects on an outdoor dot matrix display screen based on texture features provided by this embodiment. Figure 1 As shown, the method includes: S101, obtaining an original image of a pure color test screen of a dot matrix display screen; wherein the pure color test screen includes at least two pure color screens; A pure color test pattern refers to a test pattern where a dot matrix display shows a single color, typically: White screen (RGB=(255,255,255)): All channels are saturated, which is used to detect brightness anomalies (bad pixels, dark spots). Dark bad pixels appear as reduced grayscale, while bright bad pixels appear as single-channel oversaturation. Blue screen (RGB=(0,0,255)): Single channel saturation, used to detect blue sub-pixel failure. For example, when a blue sub-pixel fails, the pixel in the blue screen will appear black or other colors.

[0022] In this step, original images of at least two pure-color images are obtained. The sensitivity differences of multiple color channels are used to cover different types of defects. For example, blue bad pixels may be masked in white images, requiring special inspection of the blue image. Compared with single-color detection, the missed detection rate of the multi-color channel detection method is significantly reduced.

[0023] S102, preprocessing the original image to obtain a preprocessed image; The original image may contain interference factors, such as fluctuations in outdoor ambient light and geometric distortion caused by the camera and screen not being completely parallel. The preprocessing algorithm filters out irrelevant information and retains valid signals related to the defects, making subsequent feature extraction more accurate.

[0024] It should be noted that this embodiment does not limit the specific type of preprocessing operation used. The corresponding preprocessing means can be configured according to the interference factors existing in the original image in the actual application scenario. For example, geometric correction can be performed to eliminate optical distortion, and multi-channel noise suppression, brightness normalization, and contrast enhancement can also be performed. The details will not be repeated here.

[0025] S103, calling the Gabor filter to extract multi-scale texture features from the pre-processed image to generate multi-scale texture features; In this step, the Gabor filter is used to extract multi-scale texture features. The Gabor filter is the product of a Gaussian function and a complex exponential function. It has good localization characteristics in both the frequency domain and the spatial domain. The Gabor filter is sensitive to local texture changes (such as bad pixels and dark spots) on the dot matrix display screen, and can suppress interference from global illumination changes. In addition, by adjusting the Gabor filter parameters, texture features of specific scales and directions can be selectively amplified to adapt to different types of defects, thereby achieving accurate feature extraction for different defect types.

[0026] Multi-scale texture features refer to features with different scale parameters, including at least two scale parameters. Different scales correspond to different levels of defects. For example, small scale (2-4 pixels) can capture pixel-level details (sub-pixel arrangement, point defects) and can be used to identify single-point bad pixels and sub-pixel missing pixels; medium scale (8-16 pixels) can capture module-level texture (pixel array periodicity) and can be used to identify small areas of dark spots and uneven brightness; large scale (32-64 pixels) can capture the overall structure (box splicing, module boundaries) and can be used to identify large-area brightness differences and splicing seams.

[0027] This step constructs a filter bank covering multiple scales, including micro, small, medium, and large, to fully analyze the texture of dot matrix displays and capture defects at all levels, from pixels to modules. This technology provides feature inputs containing multi-dimensional information such as texture structure and scale for subsequent contrast and separation models, increasing the accuracy of defect detection on dot matrix displays to over 92%, making it particularly suitable for efficient and accurate detection in complex outdoor environments.

[0028] S104, based on the multi-scale texture features, separating the texture background from the defect foreground through a contrast separation model to generate a separation result; Normal areas of dot matrix displays present a regular, periodic pixel array texture that is highly predictable. Abnormal areas such as bad pixels and dark spots disrupt the regularity of the texture, manifesting as local texture mutations or abnormal enhancements. In this step, a pre-trained contrast separation model is called to construct a feature expression of the normal texture. The feature to be detected is compared with it, and the degree of deviation is calculated. If the threshold is exceeded, it is judged as a defect. The pixels in the image are divided into two categories: those that conform to the normal texture pattern (background) and those that deviate from the normal pattern (foreground).

[0029] The contrastive separation model is a pre-trained deep learning model that automatically learns features through a deep neural network, adapts to different texture patterns, and has strong generalization capabilities. In this embodiment, there is no limitation on the specific model type used for the contrastive separation model. For example, an autoencoder (AE), a variational autoencoder (VAE), a contrastive learning network, etc. can be used.

[0030] S105: Mark the defective area in the original image based on the separation result.

[0031] Based on the separation results, the defect areas in the original image are marked, converting the abstract defect information detected by the algorithm into visual and intuitive identifiers for users to view. Common marking methods include but are not limited to: color coding: filling the defect area with eye-catching colors such as red and yellow; contour drawing: outlining the defect with a border; transparency overlay: covering the defect area with a semi-transparent mask to preserve the original image details.

[0032] Based on the above introduction, this embodiment provides a texture-based defect detection method for outdoor dot-matrix displays. It obtains at least two pure-color test images and leverages the sensitivity differences between different color channels to achieve comprehensive defect detection. Furthermore, multi-scale feature extraction based on Gabor filters captures features at scales ranging from single bad pixels (microscopic), regional dark spots (mesoscopic), and module-level distortion (macroscopic), covering a wide range of defect types, from large brightness deviations to large-area brightness unevenness. This method increases the detection rate of traditionally missed scenarios, such as small bad pixels and color deviations, to over 95%. Furthermore, it exhibits strong environmental robustness, with a low false detection rate in complex lighting and vibration environments. This provides a highly reliable automatic defect detection solution for outdoor advertising and public displays.

[0033] Example 2: The above embodiment does not limit the type of pure color test image obtained. In view of the differences in physical properties of different color channels of the dot matrix display screen, this embodiment proposes to achieve collaborative analysis of white and blue images through customized Gabor filter groups and cross-channel feature fusion to improve defect detection capabilities.

[0034] Specifically, the pure color test image includes a blue image and a white image, and step S103 of calling a Gabor filter to extract multi-scale texture features from the pre-processed image can be specifically performed according to the following steps: Step S31: For the pre-processed white screen image, call the first group of Gabor filters to extract texture features as the first group of features, and generate an original feature map; Step S32: extracting texture features from the preprocessed blue screen image using a second set of Gabor filters as a second set of features; wherein the σ parameter of the second set of Gabor filters is greater than the σ parameter of the first set of Gabor filters; The σ parameter controls the width of the Gaussian envelope of the Gabor filter, directly affecting the size of the receptive field. A larger σ value increases the spatial coverage of the filter and makes it more sensitive to low-frequency texture (large-area brightness variations). A smaller σ value reduces the receptive field and focuses on high-frequency details (pixel-level defects). However, the luminous efficiency of blue dot-matrix displays is typically lower than that of red and green (approximately 60% of that of red), resulting in relatively blurred pixel textures in blue images. Increasing the σ parameter of the second filter group, for example by 20% compared to white images, expands the receptive field, compensates for signal attenuation in the blue channel, and ensures that the characteristics of blue defective pixels (such as sub-pixel failures) are effectively captured.

[0035] Step S33, calculating a channel difference response map between the first set of features and the second set of features; Normal pixels have consistent texture features in images of different colors, while defective areas (such as bad pixels and dark spots) will show characteristic differences. By calculating the channel difference response map of the first and second sets of features and using the feature contrast of different color images to highlight the defective areas, quantifying this difference can effectively identify defects that are easily missed in traditional single-channel detection.

[0036] set up and White / blue screen in the first k The response value under the Gabor filter is a channel difference response map D ( x , y ) is defined as:

[0037] in, K is the number of Gabor filters (usually K =12, corresponding to 3 scales × 4 directions), For the k The weight of each filter is determined by its sensitivity to defects (e.g., small-scale filters have higher weights). The square root of the sum of squared differences can ensure that the difference value is non-negative and amplify significant differences.

[0038] Step S34: Fusing the channel difference response map with the original feature map to generate an enhanced multi-scale texture feature as the multi-scale texture feature.

[0039] The original feature map, which refers to the Gabor features of white images, is sensitive to brightness anomalies (such as bright / dark defective pixels) and preserves high-resolution spatial detail, but has a weak response to color-specific defects (such as blue sub-pixel failures). The channel difference response map highlights feature inconsistencies between different color images and is sensitive to color deviation defects, but has low spatial localization accuracy (difference calculations lead to blurred edges) and is not very responsive to pure brightness anomalies. Combining the color sensitivity of the channel difference response map with the spatial accuracy of the original feature map achieves a dual enhancement of defect characteristics.

[0040] The multi-scale texture feature extraction method for blue and white images provided in this embodiment systematically solves the limitations of traditional single-channel detection through the collaborative Gabor feature extraction and fusion technology of dual color channels, achieving a double breakthrough in detection accuracy and robustness.

[0041] Example 3: In order to further improve the accuracy of defect detection, this embodiment proposes to introduce local binary patterns (LBP) and fuse them with multi-scale Gabor features to improve the accuracy of defect detection through the complementarity of local structural features and global texture features.

[0042] Based on the above embodiment, the following steps may be further performed: Step S106: extracting local texture features of the preprocessed image using local binary patterns; Local Binary Patterns (LBP) generate binary codes to describe local texture structures by comparing the grayscale differences between a central pixel and its neighboring pixels. LBP is highly sensitive to defects. Specifically, when the grayscale difference between a bad pixel and its neighboring pixels is significant, the LBP code exhibits a unique pattern. Furthermore, LBP is sensitive to sudden changes in pixel grayscale, accurately capturing local structural anomalies at defect boundaries.

[0043] Step S107: Fusing the local texture features with the multi-scale texture features to generate local fusion features.

[0044] The LBP feature and the Gabor feature are fused to achieve information complementarity of local structure and global texture. The fused features improve the clustering compactness of normal textures in the latent space and increase the distance between defect features and normal features. In particular, the detection rate of slight bad pixels with a color similar to the background is significantly improved.

[0045] This embodiment does not limit the feature fusion algorithm. However, an algorithm implementation is proposed herein, including: channel-wise concatenation of local texture features and multi-scale texture features to generate multi-dimensional fused features; employing an attention mechanism to calculate fusion weights; and weighted fusion of the features in the multi-dimensional fused features based on the fusion weights to form local fused features. 256-dimensional histogram features extracted from local binary patterns (LBP) are concatenated with multi-scale Gabor features (e.g., 48-dimensional) on a channel-by-channel basis to form a 304-dimensional raw fused feature vector. This adaptively focuses on defect-sensitive areas, assigning higher weights to features that contribute significantly to defect detection (e.g., LBP edge response and Gabor color difference). This can suppress environmental noise interference and reduce false alarm rates in complex outdoor environments. Of course, other feature fusion methods can also be used, and all of these can be referred to in the description of this embodiment and will not be elaborated upon here.

[0046] Then step S104 separates the texture background from the defect foreground through a contrast separation model based on the multi-scale texture features, specifically: based on the local fusion features, separates the texture background from the defect foreground through a contrast separation model.

[0047] The method provided in this embodiment utilizes the high sensitivity of LBP to local pixel anomalies. By fusing LBP features with multi-scale Gabor features, it makes up for the shortcomings of Gabor in single-point defect detection. The fusion features optimize the latent space, improve the efficiency and robustness of contrastive learning, and realize full-dimensional texture representation of microscopic details and macroscopic structures, which qualitatively improves the defect detection capability of the contrast separation model.

[0048] Example 4: In order to ensure that the optimal shooting distance and angle are maintained between the camera and the screen, so as to obtain a distortion-free, high-resolution original image, thereby meeting the high-precision requirements of pixel-level detail detection of the dot matrix display screen, step S101 obtains the original image of the pure color test screen of the dot matrix display screen. Specifically, the motor can drive the telescopic rod to move the camera to a position where the pixel-level details of the dot matrix display screen can be captured, and the lens plane is parallel to the screen surface, to capture the original image of the pure color test screen.

[0049] Defects in dot-matrix displays (such as dead pixels, uneven brightness, and pixel offset) typically occur at the micron level and require pixel-level resolution to accurately identify. A fixed camera positioned too far from the screen will result in insufficient image resolution (blurred pixel details); if the camera is too close, the viewing angle may be limited and the entire area may not be captured. A movable camera adjusts its distance to ensure that every pixel is clearly captured in the image.

[0050] Furthermore, if the camera is not parallel to the screen (at a pitch or roll angle), the image will experience perspective distortion (images appear larger when closer and smaller when farther away), leading to errors in subsequent defect detection based on texture features (such as texture extracted using Gabor filters and local binary pattern features). For example, pixels at the edge of the screen may appear stretched or compressed in a distorted image, causing the algorithm to mistakenly identify texture anomalies. The movable camera uses mechanical adjustment to ensure that the lens plane is strictly parallel to the screen, eliminating distortion and ensuring accurate feature extraction.

[0051] The original image acquisition method provided in this embodiment uses a mechatronic design to enable the camera to dynamically adjust its position and posture, which can solve the problems of insufficient resolution and angle deviation of traditional fixed cameras when shooting at long distances.

[0052] Embodiment 5: In dot matrix display screen defect detection, the marking process based on the separation results is a key step in converting the abstract defect information detected by the algorithm into a visual identifier. Through morphological optimization, geometric analysis, feature classification, and visual coding, the precise location and type of defects can be achieved. To facilitate the user's intuitive information acquisition, this embodiment proposes a method for marking defect areas. Specifically, step S105, marking the defect areas in the original image based on the separation results, can be performed according to the following steps: Step S51: performing a morphological opening operation on the separation result to obtain a first processed image; The separation result refers to the binary result obtained by separating the defect foreground from the texture background in the image through the comparative separation model, with the foreground representing the defect and the background representing the normal area. The separation result may contain some small noise points or false positives—points / regions that are not defects but are mistakenly identified as such. In this embodiment, a morphological opening operation is further used to smooth the boundaries of the foreground area, remove small foreground noise points (false positives), and retain the larger true defect area. The specific calculation process for performing the morphological opening operation on the separation result can be referred to related technologies and will not be detailed here.

[0053] Step S52: removing the false detection points whose areas are smaller than the threshold in the first processed image to obtain a second processed image; After the opening operation, there may still be some small foreground areas in the image. By calculating the area of each connected domain (continuous foreground pixel block) and setting an area threshold, areas with an area smaller than the threshold are directly deleted (reset to background) to further filter out false positives.

[0054] Step S53: Calculating geometric features of defects in the second processed image based on connected component analysis; By analyzing the connected defect areas in the image, quantitative indicators that can describe their physical properties such as shape, size, and location are extracted. The specific indicators of the calculated geometric features are not limited and can be set according to actual marking needs, so they will not be detailed here.

[0055] Step S54: classify the defects according to the geometric features and the multi-scale texture features to determine the defect type; By comprehensively utilizing the shape information (such as size and aspect ratio) and texture pattern (such as brightness changes and frequency characteristics) of the defects, the detected defects are automatically classified into different types, such as dead pixels, dark spots, scratches, etc.

[0056] Step S54: Draw a marking frame of different colors on the original image according to the defect type.

[0057] The algorithm detection results are converted into intuitive visual information, and through color coding and graphic marking, operators can quickly identify the type and severity of defects.

[0058] The defect area marking method provided in this embodiment first removes edge burrs and isolated noise points through morphological opening operations, optimizes the defect outline, and then combines area threshold filtering to significantly reduce the false detection rate of pseudo-defects such as dust reflections and random noise, ensuring that subsequent analysis focuses on real defects. Differentiated color coding is then designed according to the defect type, and an external rectangular box and type label are superimposed to improve the efficiency of manual re-inspection. The marking results can be accurately mapped to the original image coordinates, directly guiding maintenance personnel to locate the faulty module and realize the visualization of the defect distribution of the screen cluster. This method systematically improves the accuracy and engineering practicality of dot matrix display screen defect detection through a closed-loop process of denoising optimization-feature quantization-intelligent classification-visual coding.

[0059] Example 6: After long-term use, the aging risk of outdoor dot matrix displays increases dramatically. To improve the timeliness and economy of equipment maintenance, this embodiment proposes a defect risk warning method for dot matrix displays. Specifically, after marking the defect area in the original image based on the separation result in step S105, the following steps can be further performed: Step S108: Acquire defect area data detected in history as historical data; The system stores information such as defect location, area, and type from past inspections (e.g., results from the past 10 days and 20 inspections) to generate time series data. For example, if a dot matrix display screen detects a 20-pixel dark spot on day 1 and a 35-pixel dark spot on day 3, this data will be archived as historical data for subsequent trend analysis.

[0060] Step S109: Compare and analyze the historical data with the current defect area to calculate the defect diffusion speed; Compare the currently detected defect area with historical data and calculate its rate of change over time. Taking area as an example, if a defect in historical data expands from 100 pixels to 200 pixels over three days, with a time interval of 72 hours, the diffusion rate is (200 - 100) / 72 ≈ 1.39 pixels / hour. Furthermore, the diffusion rate can also refer to changes in defect shape (such as the growth rate of the major axis) or positional migration (such as the movement trajectory within the screen pixel array), which is not limited in this embodiment.

[0061] Step S110: If the defect diffusion speed exceeds the diffusion threshold, output a warning message.

[0062] The system has a preset diffusion threshold (e.g., 2 pixels / hour). When the calculated diffusion rate exceeds this threshold, it indicates that the defect is accelerating (e.g., a pixel failure on a screen panel is spreading from a single point to the surrounding area). At this point, the system issues a warning (e.g., audible and visual alarms, generating a maintenance work order), prompting maintenance personnel to intervene before the defect expands and causes equipment failure.

[0063] The method provided in this embodiment upgrades passive detection to active prevention. Through time series analysis of historical data and current detection results, it can predict the development trend of defects. Early warning can help the production line replace components in advance and reduce scrap rates.

[0064] Embodiment seven: Further references Figure 2 , which shows an exemplary structural block diagram of an outdoor dot matrix display screen defect detection device based on texture features according to an embodiment of the present application, which mainly includes: an image acquisition module, a preprocessing module, a feature extraction module, a separation processing module and a defect marking module. The outdoor dot matrix display screen defect detection device based on texture features adopts a modular design and realizes defect detection of outdoor dot matrix display screens in complex scenes through five core units.

[0065] The image acquisition module is used to acquire the original image of the pure color test screen of the dot matrix display screen; wherein the pure color test screen includes at least two pure color screens; A preprocessing module, used for preprocessing the original image to obtain a preprocessed image; The feature extraction module is used to extract multi-scale texture features from the pre-processed image by calling the Gabor filter to generate multi-scale texture features; A separation processing module is used to separate the texture background and the defect foreground based on multi-scale texture features through a comparative separation model to generate a separation result; The defect marking module is used to mark the defect area in the original image based on the separation result.

[0066] The texture feature-based outdoor dot matrix display screen defect detection device provided in this embodiment can be used in conjunction with the above method embodiment, and the repeated parts will not be repeated in this embodiment.

[0067] In the texture feature-based outdoor dot matrix display screen defect detection device provided in this embodiment, the image acquisition module adopts multi-pure color image collaborative acquisition, the preprocessing module is deeply coupled with the feature extraction module, and multi-scale texture extraction is performed through Gabor filter. The separation processing module is based on a deep learning model of the contrast learning mechanism, and complex lighting and dust noise are suppressed through self-supervised feature contrast. The false detection rate is significantly reduced, and the detection accuracy in complex outdoor environments is significantly improved, providing an engineering solution for the automated inspection of dot matrix displays.

[0068] Embodiment 8: This embodiment provides a texture feature-based outdoor dot matrix display screen defect detection system, which mainly includes: A movable camera is used to capture raw images of a pure-color test pattern on a dot matrix display; the pure-color test pattern consists of at least two pure-color images. The movable camera, driven by a motor, achieves pixel-level resolution (single-pixel ratio ≥ 2×2 pixels) and lens plane parallelism control (error ≤ 1°), overcoming the resolution limitations and distortion associated with traditional fixed cameras in long-distance photography.

[0069] A texture feature-based outdoor dot matrix display screen defect detection device is used for an original image, preprocesses the original image to obtain a preprocessed image, calls a Gabor filter to extract multi-scale texture features from the preprocessed image, generates multi-scale texture features, and based on the multi-scale texture features, separates the texture background from the defect foreground through a comparative separation model to generate a separation result; and marks the defect area in the original image based on the separation result. The device can be referred to in the above embodiment and will not be described in detail here.

[0070] The user interaction device is connected to the user device and is used to visually output the original image after the defect area is marked.

[0071] Embodiment 9: Reference below Figure 3 , Figure 3 A schematic diagram of the structure of a computer system suitable for implementing a server according to an embodiment of the present application is shown.

[0072] like Figure 3As shown, the computer system includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 302 or programs loaded from a storage unit 308 into a random access memory (RAM) 303. RAM 303 also stores various programs and data required for the system's operating instructions. CPU 301, ROM 302, and RAM 303 are connected to each other via a bus 304. An input / output (I / O) interface 305 is also connected to bus 304.

[0073] The following components are connected to the I / O interface 305: an input section 306 including a keyboard, mouse, and the like; an output section 307 including devices such as a cathode ray tube (CRT), a liquid crystal display (LCD), and speakers; a storage section 308 including devices such as a hard disk; and a communication section 309 including a network interface card such as a LAN card or a modem. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. Removable media 311, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, is installed in the drive 310 as needed, so that computer programs read from the media can be installed in the storage section 308 as needed.

[0074] In particular, according to the embodiment of the present application, the above reference flow chart Figure 1 The described processes can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for executing the method illustrated in the flowchart. In such an embodiment, the computer program contains program code for executing the method illustrated in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 309 and / or installed from removable media 311. When the computer program is executed by the central processing unit (CPU) 301, the aforementioned functions defined in the system of the present application are performed.

[0075] It should be noted that the computer-readable medium described herein may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may, for example, be an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection having one or more conductors, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. This propagated data signal may take a variety of forms, including, but not limited to, electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transfer a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium may be transmitted using any suitable medium, including but not limited to wireless, wire, optical cable, RF, or any suitable combination thereof.

[0076] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operating instructions of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the aforementioned module, program segment, or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than the order marked in the accompanying drawings. For example, the boxes represented by two connections can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of the boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operating instruction, or can be implemented using a combination of dedicated hardware and computer instructions.

[0077] The units or modules involved in the embodiments described in this application may be implemented in software or hardware. The units or modules described may also be provided in a processor. The names of these units or modules do not, in certain circumstances, constitute limitations on the units or modules themselves.

[0078] The above description is merely a preferred embodiment of the present application and an illustration of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this application is not limited to technical solutions formed by a specific combination of the aforementioned technical features. It also encompasses other technical solutions formed by any combination of the aforementioned technical features or their equivalents, without departing from the aforementioned disclosed concepts. For example, a technical solution formed by replacing the aforementioned features with (but not limited to) technical features with similar functions disclosed in this application.

Claims

1. A method for detecting defects on an outdoor dot matrix display screen based on texture features, characterized in that: include: Acquire an original image of a pure color test screen of a dot matrix display screen; wherein the pure color test screen includes at least two pure color screens; Preprocessing the original image to obtain a preprocessed image; Calling a Gabor filter to perform multi-scale texture feature extraction on the pre-processed image to generate multi-scale texture features; Based on the multi-scale texture features, the texture background and the defect foreground are separated by a contrast separation model to generate a separation result; The defective area in the original image is marked based on the separation result.

2. The method according to claim 1, wherein The pure color test picture includes a blue picture and a white picture, and the calling of the Gabor filter to perform multi-scale texture feature extraction on the pre-processed image includes: For the preprocessed white screen image, the first set of Gabor filters is called to extract texture features as the first set of features and generate the original feature map; For the preprocessed blue screen image, calling a second group of Gabor filters to extract texture features as a second group of features; wherein the σ parameter of the second group of Gabor filters is greater than the σ parameter of the first group of Gabor filters; calculating a channel difference response map between the first set of features and the second set of features; The channel difference response map is fused with the original feature map to generate an enhanced multi-scale texture feature as the multi-scale texture feature.

3. The method according to claim 1, wherein Also includes: Extracting local texture features of the preprocessed image using local binary patterns; Fusing the local texture feature with the multi-scale texture feature to generate a local fusion feature; Based on the multi-scale texture features, the texture background and the defect foreground are separated by a contrast separation model. Specifically, based on the local fusion features, the texture background and the defect foreground are separated by a contrast separation model.

4. The method according to claim 3, wherein The fusing of the local texture feature with the multi-scale texture feature comprises: Cascading the local texture features and the multi-scale texture features by channel to generate a multi-dimensional fusion feature; Use attention mechanism to calculate fusion weights; The features in the multi-dimensional fusion feature are weightedly fused based on the fusion weight to serve as the local fusion feature.

5. The method according to claim 1, wherein The method of obtaining the original image of the pure color test screen of the dot matrix display screen includes: The motor drives the telescopic rod to move the camera to a position where it can capture pixel-level details of the dot matrix display screen, and the lens plane is parallel to the screen surface, thereby capturing the original image of the pure color test screen.

6. The method according to claim 1, wherein The marking of the defective area in the original image based on the separation result includes: performing a morphological opening operation on the separation result to obtain a first processed image; removing falsely detected points whose areas are smaller than a threshold value from the first processed image to obtain a second processed image; Calculating geometric features of defects in the second processed image based on connected component analysis; Classify the defects according to the geometric features and the multi-scale texture features to determine the defect type; Marking boxes of different colors are drawn on the original image according to the defect type.

7. The method according to claim 1, wherein After marking the defective area in the original image based on the separation result, the method further includes: Obtaining defect area data from historical inspections as historical data; Comparing and analyzing the historical data with the current defect area, and calculating the defect diffusion rate; If the defect diffusion speed exceeds the diffusion threshold, a warning message is output.

8. A device for detecting defects on an outdoor dot matrix display screen based on texture features, characterized in that: include: An image acquisition module is used to acquire an original image of a pure color test screen of a dot matrix display screen; wherein the pure color test screen includes at least two pure color screens; A preprocessing module, configured to preprocess the original image to obtain a preprocessed image; A feature extraction module is used to extract multi-scale texture features from the pre-processed image by calling a Gabor filter to generate multi-scale texture features; A separation processing module, configured to separate the texture background from the defect foreground based on the multi-scale texture features by comparing the separation model and generating a separation result; A defect marking module is used to mark the defect area in the original image based on the separation result.

9. A texture feature-based outdoor dot matrix display screen defect detection system, characterized in that: include: A movable camera is used to capture an original image of a pure color test screen of a dot matrix display screen; wherein the pure color test screen includes at least two pure color screens; The texture feature-based outdoor dot matrix display screen defect detection device according to claim 8 is used for the original image, preprocessing the original image to obtain a preprocessed image, calling a Gabor filter to extract multi-scale texture features from the preprocessed image to generate multi-scale texture features, based on the multi-scale texture features, separating the texture background from the defect foreground through a comparative separation model to generate a separation result; and marking the defect area in the original image based on the separation result; The user interaction device is used to visually output the original image after marking the defect area.

10. A server comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the method according to any one of claims 1 to 7 are implemented.

Images