Anti-motion blur video stream edge detection system

Abstract

This invention discloses an anti-motion blur video stream edge detection system, belonging to the field of computer vision technology. Utilizing motion blur analysis and adaptive filtering techniques, this invention accurately identifies and corrects motion blur in dynamic scenes, avoiding the destruction of edge details during deblurring as is common in traditional methods. It can generate clear and complete edge images in complex environments. Through real-time processing requirement analysis and dynamic optimization mechanisms, the system automatically adjusts filtering parameters and feature extraction complexity, ensuring high detection accuracy while meeting real-time requirements. This significantly improves the overall performance and practicality of the system, enabling it to perform excellently in applications with high real-time requirements, such as intelligent monitoring and autonomous driving, ensuring efficient and stable operation in real-world scenarios.

Description

Technical Field

[0001] This invention relates to the field of computer vision technology, specifically to a video stream edge detection system that resists motion blur. Background Technology

[0002] In the field of modern visual technology, video stream edge detection is a key technology with irreplaceable value in applications such as intelligent surveillance, autonomous driving, and virtual reality. It is not only the foundation of image understanding but also the core support for dynamic scene analysis. However, despite some progress in related research, existing methods still show significant shortcomings when facing complex dynamic environments. Many solutions often struggle to balance real-time performance and accuracy, especially when dealing with scenes of high-speed motion or drastic changes in lighting. They are prone to edge detection results being broken or blurred due to neglecting the dynamic correlation between inter-frame information, which in turn affects the reliability of subsequent tasks.

[0003] A deeper analysis of the challenges in this field reveals that motion blur in dynamic scenes is the primary obstacle. Motion blur not only obscures edge details but also exacerbates information loss due to the continuity differences between video frames. More importantly, this blurring phenomenon is closely related to the complexity of motion trajectories between frames. When motion trajectories cannot be accurately captured and modeled, traditional filtering methods often destroy the integrity of edges while deblurring, resulting in loss of details or misjudgment. This problem has further evolved into how to dynamically adjust filtering strategies while deeply mining the correlation patterns of features between frames in order to achieve accurate repair and edge enhancement of blurred areas. Summary of the Invention

[0004] The purpose of this invention is to provide a video stream edge detection system that resists motion blur. Through edge continuity analysis, motion blur repair, and real-time optimization, the system significantly improves the integrity and accuracy of edge detection in dynamic scenes.

[0005] The objective of this invention can be achieved through the following technical solutions: This application provides a video stream edge detection system that resists motion blur, including: The video stream segmentation module obtains the original video data stream from the continuous frame sequence through the video stream data acquisition module, performs preliminary segmentation processing on the high-speed motion features in the dynamic scene, and obtains the segmented video segment sequence. The motion blur analysis module, based on the segmented video segment sequence, uses a motion estimation algorithm to analyze the displacement vectors between adjacent frames, makes a preliminary judgment on the degree of motion blur based on the magnitude of the change in displacement vectors, and determines the distribution of the blur-affected area. The adaptive filtering and deblurring module obtains an adaptive filtering parameter adjustment model based on the distribution of the blur-affected region, extracts the filter kernel parameters that match the degree of motion blur, and obtains a preliminary deblurred frame image after targeted filtering processing. The edge continuity analysis module analyzes the grayscale variation patterns of pixels between frames through preliminary deblurred frame images, and obtains edge continuity features between adjacent frames by combining the inter-frame feature association mapping method to determine potential edge break regions. The edge restoration and enhancement module extracts supplementary information from the continuous features of adjacent frames for potential edge breakage areas, and uses deep feature fusion technology to perform pixel-level restoration of the breakage areas to obtain the restored edge-enhanced image. The illumination correction module analyzes the degree of interference of illumination changes on edge detection based on the restored edge enhancement image, and obtains the illumination-corrected edge image. The multi-scale edge extraction and optimization module takes the edge image after illumination correction and combines it with the complex environmental features in the dynamic scene to obtain a multi-scale edge detection operator for secondary edge extraction, judges whether the edge details are complete, and obtains the final edge detection result image.

[0006] Furthermore, the video stream segmentation processing module includes an inter-frame difference analysis unit, a motion trajectory tracking unit, and an optical flow analysis and secondary segmentation unit; The inter-frame difference analysis unit is used to capture the input continuous frame data in real time, analyze the continuous frames using the inter-frame difference method, determine the changing areas in the dynamic scene, and obtain the preliminary feature areas of high-speed motion. When the motion trajectory tracking unit detects a preliminary feature region of high-speed motion, it tracks the motion trajectory of that region to determine the continuity characteristics of the moving object. Then, it combines a preset threshold to segment the video data stream on the time axis to obtain preliminary segmented video clips. The optical flow analysis and secondary segmentation unit uses optical flow analysis to extract the high-speed motion characteristics of each segment in the initial segmentation, and obtains the motion intensity distribution of each segment. When the motion intensity of a segment exceeds a preset threshold, the segment is further segmented to determine the final video segment sequence.

[0007] Furthermore, the motion fuzzy analysis module includes a dense optical flow calculation unit, a fuzzy region extraction unit, a fuzziness degree classification unit, and a fuzzy region spatial description unit; The dense optical flow calculation unit acquires adjacent frame pairs, uses the dense optical flow algorithm to calculate the displacement vector field of each pair of adjacent frames, obtains the inter-frame pixel-level displacement vector, calculates the amplitude of each pixel displacement vector, and generates an amplitude distribution map. The fuzzy region extraction unit determines that there is significant motion in a region when the amplitude of a certain region in the amplitude distribution map exceeds a preset threshold. It then obtains the coordinates of the high-amplitude region to obtain the potential fuzzy region and uses Gaussian filtering to smooth the displacement vector field to generate a smoothed vector field. The fuzziness level classification unit determines the boundary distribution of the fuzzy region, extracts the amplitude change trend from the smoothed vector field, and uses the K-means clustering algorithm to classify the amplitude change trend to obtain the fuzziness level distribution. The fuzzy region spatial description unit generates a spatial mask for the fuzzy region based on the distribution of the degree of fuzziness, determines the final distribution of the fuzzy influence region, performs grid division, generates statistical features of the region distribution, and obtains a spatial quantitative description of the fuzzy region.

[0008] Furthermore, the adaptive filtering and deblurring module includes a blurred region analysis unit, a filter parameter matching unit, an adaptive filtering and optimization unit, and a local detail restoration unit; The fuzzy region analysis unit analyzes the fuzzy effects and regional distribution characteristics of the input image to obtain a distribution map of the fuzzy region, determine the range of fuzzy effects, and then, in combination with the relevant characteristics of motion fuzziness and fuzziness degree, judge the distribution pattern of the fuzzy type. When the filter parameter matching unit detects that motion blur occupies the main area, it uses a preset motion trajectory analysis tool to extract the motion direction and intensity, and then uses a pre-established adjustment model to match the filter kernel value corresponding to the degree of blur, and determines the appropriate set of filter kernel parameters. The adaptive filtering and optimization unit performs adaptive filtering on the blurred region in accordance with business requirements, generates an initial frame image with preliminary deblurring, analyzes the residual blurred region after image processing, and performs secondary processing by iteratively adjusting the filter kernel value when the residual blurred region exceeds a preset threshold, to obtain the optimized deblurred frame image. The local detail restoration unit acquires the overall sharpness distribution of the image processing, determines whether there is local blurring, determines the final sharp frame image, and then verifies it using an image quality assessment tool in conjunction with the overall deblurring effect to obtain an output image that meets business requirements.

[0009] Furthermore, the edge continuity analysis module includes an image enhancement and grayscale analysis unit, an inter-frame feature correlation extraction unit, and a break region localization unit; The image enhancement and grayscale analysis unit performs preliminary processing on the blurred frame image, uses image enhancement technology for noise reduction and contrast adjustment, and then scans the pixel grayscale information frame by frame to obtain the grayscale change pattern between frames and determine the areas with significant grayscale changes. The inter-frame feature association extraction unit extracts relevant features between adjacent frames by combining inter-frame feature mapping methods for regions with significant grayscale changes. When there are significantly discontinuous feature values ​​in the inter-frame feature association matrix, edge continuity analysis is performed on the corresponding region to determine the possibility of potential edge breaks. The fracture region localization unit, based on the edge continuity analysis results, uses an edge detection algorithm such as the Canny algorithm to locate potential fracture regions, obtains a preliminary distribution map of the fracture regions, and combines the contextual features between adjacent frames to obtain the precise boundary of the fracture regions and determine the final fracture region range.

[0010] Furthermore, the edge restoration and enhancement module includes a feature extraction and region localization unit, a feature fusion and pixel restoration unit, and an edge enhancement and image fusion unit; The feature extraction and region localization unit obtains inter-frame information from adjacent frame groups, analyzes continuous features using a preset feature extraction method, locates potential regions, and uses image processing technology to mark edge breakage positions to determine the boundary range of the breakage region. The feature fusion and pixel repair unit, when the boundary range of the labeled fracture region is significantly different from the continuous features of the adjacent frame group, integrates the inter-frame information and fracture region features through a deep fusion method, and then uses a convolutional neural network to perform pixel-level processing on the fused feature set to perform pixel repair operation on the edge fracture region. The edge enhancement and image fusion unit uses edge enhancement technology to optimize the boundary of the fractured area, obtain enhanced edge detail information, and performs overall image processing to fuse the global features of the repaired image and generate the final edge-enhanced repaired image.

[0011] Furthermore, the illumination correction module includes an illumination distribution feature extraction unit, a local contrast adjustment unit, and an illumination normalization and optimization unit; The illumination distribution feature extraction unit extracts illumination distribution features from the edge enhancement image, calculates illumination change values ​​using a histogram equalization method, and adjusts the edge enhancement image using a local contrast enhancement algorithm when the illumination change data exceeds a preset threshold range to obtain a contrast-corrected image. The local contrast adjustment unit extracts edge features from the contrast-corrected image, uses the Canny edge detection algorithm, and performs smoothing processing using Gaussian filtering based on the noise distribution in the edge image to obtain a smoothed edge image.

[0012] Furthermore, the illumination normalization and optimization unit extracts illumination correction parameters from the smooth edge image, performs illumination normalization processing using an adaptive histogram equalization method, optimizes edge continuity using morphological processing methods based on the edge intensity of the illumination correction image, extracts the final edge features, and calculates the edge gradient using the Sobel operator to obtain the final edge image.

[0013] Furthermore, the multi-scale edge extraction and optimization module includes an image preprocessing unit, a multi-scale edge detection unit, and an edge detection result optimization unit; The image preprocessing unit preprocesses the edge image after illumination correction and removes noise interference using image smoothing technology to obtain preliminary processed edge image data. The multi-scale edge detection unit acquires environmental feature information in dynamic scenes, applies multi-scale detection methods to extract edge features at different scales, and uses edge operators to perform secondary extraction operations to obtain more refined edge line information, thus obtaining edge detail data after secondary extraction. If there are breaks or missing phenomena in the edge detail data after secondary extraction, the broken parts are repaired through edge connection technology to obtain repaired edge detail information. The edge detection result optimization unit analyzes the edge integrity, obtains integrity evaluation indicators, determines whether the edge details meet the preset integrity standards, and then, in combination with the overall features of the detected image, adjusts the display parameters of the edge details to obtain the final edge detection result image. If there are still local blurred areas in the final edge detection result image, local sharpening technology is used to enhance the blurred areas.

[0014] Furthermore, after obtaining the final edge detection result image, the process also includes: analyzing real-time processing requirements; if the processing time exceeds the preset threshold range, adjusting the filtering parameters and the computational complexity of feature extraction, and determining the optimized processing flow configuration.

[0015] The beneficial effects of this invention are as follows: This invention effectively solves the problem of edge detection being easily broken and blurry in dynamic scenes by using edge continuity analysis and repair enhancement methods. In complex dynamic environments, it can accurately locate and repair potential edge breakage areas, significantly improving the integrity and accuracy of edge detection, providing a high-quality image foundation for subsequent tasks, and greatly enhancing the reliability and accuracy of the system in target recognition, tracking and other applications. By combining motion blur analysis and adaptive filtering techniques, motion blur problems in dynamic scenes are accurately identified and repaired. By dynamically adjusting the filtering strategy, the destruction of edge details during deblurring by traditional methods is avoided. Clear and complete edge images can be generated in scenes with high-speed motion or drastic changes in lighting, significantly improving the adaptability and reliability of edge detection in complex dynamic environments and expanding its application scope in fields such as intelligent monitoring and autonomous driving. By combining real-time processing requirements analysis with dynamic optimization mechanisms, the system effectively balances the conflict between real-time performance and accuracy. When dealing with complex dynamic scenarios, the system can automatically adjust filtering parameters and feature extraction complexity to ensure that while meeting real-time requirements, it maintains high detection accuracy. This optimization mechanism enables the edge detection system to perform well in applications with high real-time requirements, such as intelligent monitoring and autonomous driving, significantly improving the overall performance and practicality of the system and ensuring its efficient and stable operation in real-world scenarios. Attached Figure Description

[0016] To better understand and implement this application, the technical solution is described in detail below with reference to the accompanying drawings.

[0017] Figure 1 A schematic diagram of the anti-motion blur video stream edge detection system provided in this application; Figure 2 A schematic diagram of the adaptive filtering and deblurring module in the anti-motion blurring video stream edge detection system provided in this application; Figure 3 This is a flowchart illustrating the multi-scale edge extraction and optimization module in the anti-motion blur video stream edge detection system provided in this application. Detailed Implementation

[0018] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, exemplary embodiments will be described in detail below, examples of which are illustrated in the accompanying drawings. In the following description relating to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of methods and systems consistent with some aspects of this application as detailed in the appended claims.

[0019] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0020] The following detailed description of the specific implementation methods, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided in detail.

[0021] Please see Figures 1-3 This embodiment provides a video stream edge detection system that resists motion blur, including: The video stream segmentation module acquires the raw video data stream from the continuous frame sequence through the video stream data acquisition module, performs preliminary segmentation processing on the high-speed motion features in the dynamic scene, and obtains the segmented video segment sequence.

[0022] Furthermore, the video stream segmentation processing module includes an inter-frame difference analysis unit, a motion trajectory tracking unit, and an optical flow analysis and secondary segmentation unit; The inter-frame difference analysis unit is used to capture the input continuous frame data in real time, obtain the original video data stream, analyze the continuous frames using the inter-frame difference method, determine the changing areas in the dynamic scene, and obtain the preliminary feature areas of high-speed motion. When the motion trajectory tracking unit detects a preliminary feature region of high-speed motion, it tracks the motion trajectory of that region to determine the continuity characteristics of the moving object. Then, it combines a preset threshold to segment the video data stream on the time axis to obtain preliminary segmented video clips. The optical flow analysis and secondary segmentation unit uses optical flow analysis to refine and extract the high-speed motion characteristics within each segment of the initial segmentation, obtaining the motion intensity distribution of each segment. When the motion intensity of a segment exceeds a preset threshold, the segment is further segmented to determine the final video segment sequence and generate a structured data flow record, completing the entire process of segmentation.

[0023] Specifically, the video stream segmentation module can efficiently and accurately capture high-speed motion features in dynamic scenes from continuous frame sequences and generate segmented video segment sequences. This module effectively solves the problem of accurately capturing high-speed moving targets in dynamic scenes, avoiding the problems of inaccurate edge detection or information loss caused by the inability to accurately segment motion regions in traditional methods. Through the synergistic effect of inter-frame difference analysis, motion trajectory tracking, and optical flow analysis, the system can quickly locate changing regions, track the continuous features of moving objects, and perform secondary segmentation based on motion intensity to generate structured data flow records. This not only improves the efficiency and accuracy of video stream processing but also provides a clear and orderly video segment foundation for subsequent motion blur analysis and edge detection, significantly improving the adaptability and reliability of the entire system in complex dynamic environments.

[0024] The motion blur analysis module analyzes the displacement vectors between adjacent frames using a motion estimation algorithm based on the segmented video segment sequence. It makes a preliminary judgment on the degree of motion blur based on the magnitude of the change in displacement vectors and determines the distribution of the blur-affected area.

[0025] Furthermore, the motion fuzzy analysis module includes a dense optical flow calculation unit, a fuzzy region extraction unit, a fuzziness degree classification unit, and a fuzzy region spatial description unit; The dense optical flow calculation unit obtains adjacent frame pairs from the video segment sequence, calculates the displacement vector field of each pair of adjacent frames using the dense optical flow algorithm, obtains the inter-frame pixel-level displacement vector, calculates the amplitude of each pixel displacement vector, generates an amplitude distribution map, and determines the spatial characteristics of amplitude changes. The fuzzy region extraction unit determines that there is significant motion in a region when the amplitude of a certain region in the amplitude distribution map exceeds a preset threshold. It then obtains the coordinates of the high-amplitude region to obtain the potential fuzzy region and uses Gaussian filtering to smooth the displacement vector field to generate a smoothed vector field. The fuzziness level classification unit determines the boundary distribution of the fuzzy region, extracts the amplitude change trend from the smoothed vector field, and uses the K-means clustering algorithm to classify the amplitude change trend to obtain the fuzziness level distribution. The fuzzy region spatial description unit generates a spatial mask for the fuzzy region based on the distribution of the degree of fuzziness, determines the final distribution of the fuzzy influence region, performs grid division, generates statistical features of the region distribution, and obtains a spatial quantitative description of the fuzzy region.

[0026] Specifically, the motion blur analysis module can accurately identify and quantify motion-blurred regions in dynamic scenes, effectively solving the problem that traditional methods struggle to accurately determine the degree and extent of blur. This module utilizes techniques such as dense optical flow calculation, blur region extraction, blur degree classification, and spatial description to analyze the displacement vectors of adjacent frames from a segmented video sequence, generating amplitude distribution maps, extracting potential blur regions, and classifying and spatially quantifying the blur degree. This not only allows for rapid location of blur regions but also precise assessment of blur degree, providing detailed and accurate region information for subsequent deblurring processing. This significantly improves the accuracy and reliability of edge detection in complex dynamic environments, ensuring the clarity and integrity of the final output image.

[0027] The adaptive filtering and deblurring module obtains an adaptive filtering parameter adjustment model based on the distribution of the blur-affected region, extracts the filter kernel parameters that match the degree of motion blur, and obtains a preliminary deblurred frame image after targeted filtering processing.

[0028] Furthermore, the adaptive filtering and deblurring module includes a blurred region analysis unit, a filter parameter matching unit, an adaptive filtering and optimization unit, and a local detail restoration unit; The fuzzy region analysis unit analyzes the fuzzy effects and regional distribution characteristics of the input image to obtain a distribution map of the fuzzy region, determine the preliminary fuzzy influence range, and, based on the regional distribution information in the distribution map and the correlation characteristics of motion fuzziness and fuzziness degree, determine the distribution pattern of the fuzzy type. When the filter parameter matching unit detects that motion blur occupies the main area, it uses a preset motion trajectory analysis tool to extract the motion direction and intensity, obtains the quantitative distribution data of the blur degree, then obtains the adaptive filter parameters, and uses a pre-established adjustment model to match the filter kernel value corresponding to the blur degree to determine the appropriate filter kernel parameter set. The adaptive filtering and optimization unit, in conjunction with the business requirements of the processing, performs adaptive filtering on the blurred region to generate an initial frame image with preliminary deblurring. Based on the deblurring effect of the initial frame image, it analyzes the residual blurred region after image processing. When the residual blurred region exceeds a preset threshold, it performs secondary processing by iteratively adjusting the filter kernel value to obtain an optimized deblurred frame image. The local detail restoration unit acquires the overall sharpness distribution of the image processing, determines whether there is local blurring, and if there is local blurring, performs detail restoration using a local region enhancement tool to determine the final sharp frame image. Based on the detail features of the final sharp frame image and the overall effect of deblurring, the image quality assessment tool is used for verification to obtain an output image that meets business requirements.

[0029] Specifically, the adaptive filtering and deblurring module efficiently filters areas affected by motion blur, significantly improving image clarity and detail integrity. This module accurately identifies blurred regions and dynamically adjusts filter kernel parameters to match regions with varying degrees of blur. Combined with motion trajectory analysis and image quality assessment tools, the module effectively removes blur while preserving edge details, ensuring the final output image meets business requirements. This solves the problem of traditional methods easily losing details or making misjudgments during deblurring, significantly improving the accuracy and reliability of edge detection.

[0030] The edge continuity analysis module analyzes the grayscale variation patterns of pixels between frames by initially deblurring the frame image, and obtains the edge continuity features between adjacent frames by combining the inter-frame feature association mapping method, thus identifying potential edge break regions.

[0031] Furthermore, the edge continuity analysis module includes an image enhancement and grayscale analysis unit, an inter-frame feature correlation extraction unit, and a break region localization unit; The image enhancement and grayscale analysis unit performs preliminary processing on the blurred frame image, uses image enhancement technology to denoise and adjust the contrast of the original data to obtain the processed first image dataset, scans the pixel grayscale information frame by frame to obtain the grayscale change pattern between frames, and determines the areas with significant grayscale changes. The inter-frame feature association extraction unit extracts relevant features between adjacent frames by combining inter-frame feature mapping methods for regions with significant grayscale changes, and obtains an inter-frame feature association matrix. When there are significantly discontinuous feature values ​​in the inter-frame feature association matrix, edge continuity analysis is performed on the corresponding region to determine the possibility of potential edge breakage. The fracture region localization unit, based on the edge continuity analysis results, uses an edge detection algorithm such as the Canny algorithm to locate potential fracture regions, obtains a preliminary distribution map of the fracture regions, performs further image analysis, and combines the contextual features between adjacent frames to obtain the precise boundary of the fracture region and determine the final fracture region range.

[0032] Specifically, the edge continuity analysis module can accurately locate and identify potential edge breakage areas, effectively solving the edge breakage problem caused by motion blur in dynamic scenes. This module utilizes image enhancement and grayscale analysis, inter-frame feature correlation extraction, and breakage area localization techniques to scan and analyze the initially deblurred frame images frame by frame, extracting inter-frame related features and constructing a feature correlation matrix. By judging the discontinuity of feature values, combined with edge detection algorithms (such as the Canny algorithm) and contextual feature analysis, the module can accurately determine the range of the breakage area. This not only effectively repairs edge breaks but also enhances the integrity and continuity of edge detection, significantly improving the system's robustness and reliability in complex dynamic environments, providing a solid foundation for subsequent edge repair and enhancement.

[0033] The edge restoration and enhancement module extracts supplementary information from the continuity features of adjacent frames for potential edge breakage areas, and uses deep feature fusion technology to perform pixel-level restoration of the breakage areas, resulting in a restored edge-enhanced image.

[0034] Furthermore, the edge restoration and enhancement module includes a feature extraction and region localization unit, a feature fusion and pixel restoration unit, and an edge enhancement and image fusion unit; The feature extraction and region localization unit obtains inter-frame information from adjacent frame groups, analyzes continuous features using a preset feature extraction method to obtain preliminary feature mapping data, performs region localization on potential regions, and uses image processing technology to mark edge breakage positions to determine the boundary range of the breakage region. When the boundary range of the labeled fractured region differs significantly from the continuous features of adjacent frame groups, the feature fusion and pixel repair unit integrates the inter-frame information and fractured region features through a deep fusion method to obtain a fused feature set. Then, a convolutional neural network is used to process the fused feature set at the pixel level, and pixel repair operation is performed on the edge fractured region to generate preliminary repaired image data. The edge enhancement and image fusion unit uses edge enhancement technology to optimize the boundary of the fractured area, obtain enhanced edge detail information, and performs overall image processing to fuse the global features of the repaired image and generate the final edge-enhanced repaired image.

[0035] Specifically, the edge repair and enhancement module can efficiently repair potential edge breakage areas, significantly improving the integrity and clarity of edge detection. This module utilizes techniques such as feature extraction and region localization, feature fusion and pixel repair, and edge enhancement and image fusion to not only effectively solve the edge breakage problem but also enhance the clarity and continuity of edge details. This significantly improves the robustness and reliability of edge detection in complex dynamic environments, providing a high-quality image foundation for subsequent image analysis and processing.

[0036] The illumination correction module analyzes the degree of interference of illumination changes on edge detection based on the repaired edge enhancement image. When the detected illumination change exceeds the preset threshold range, illumination normalization processing is performed through local contrast adjustment method to obtain the illumination-corrected edge image.

[0037] Furthermore, the illumination correction module includes an illumination distribution feature extraction unit, a local contrast adjustment unit, and an illumination normalization and optimization unit; The illumination distribution feature extraction unit extracts illumination distribution features from the edge enhancement image, calculates illumination change values ​​using a histogram equalization method, and obtains illumination change data. When the illumination change data exceeds a preset threshold range, a local contrast enhancement algorithm is used to adjust the edge enhancement image to obtain a contrast-corrected image. The local contrast adjustment unit extracts edge features from the contrast-corrected image, uses the Canny edge detection algorithm to generate an initial edge image, and uses a Gaussian filtering method to smooth the edge image based on the noise distribution in the initial edge image, thereby obtaining a smooth edge image. The illumination normalization and optimization unit extracts illumination correction parameters from the smooth edge image, performs illumination normalization processing using an adaptive histogram equalization method to obtain an illumination correction image, optimizes edge continuity using morphological processing methods based on the edge intensity of the illumination correction image to obtain an optimized edge image, extracts final edge features, calculates edge gradients using the Sobel operator, and obtains the final edge image.

[0038] Specifically, the illumination correction module significantly reduces the interference of illumination changes on edge detection, improving its stability and accuracy. This module accurately analyzes and corrects illumination changes through techniques such as illumination distribution feature extraction, local contrast adjustment, and illumination normalization. When an illumination change exceeds a preset threshold, the system automatically adjusts local contrast, smooths noise, and optimizes edge continuity. The resulting illumination-corrected image is not only clearer but also has more prominent edge features, providing a high-quality image foundation for subsequent edge detection and analysis, significantly enhancing the system's robustness and reliability under complex lighting conditions.

[0039] The multi-scale edge extraction and optimization module takes the edge image after illumination correction and combines it with the complex environmental features in the dynamic scene to obtain a multi-scale edge detection operator for secondary edge extraction, judges whether the edge details are complete, and obtains the final edge detection result image.

[0040] Furthermore, the multi-scale edge extraction and optimization module includes an image preprocessing unit, a multi-scale edge detection unit, and an edge detection result optimization unit; The image preprocessing unit preprocesses the edge image after illumination correction and removes noise interference using image smoothing technology to obtain preliminary processed edge image data. The multi-scale edge detection unit acquires environmental feature information in dynamic scenes, applies multi-scale detection methods to extract edge features at different scales, determines the multi-scale edge feature set, and uses edge operators to perform secondary extraction operations to obtain more refined edge line information, thus obtaining edge detail data after secondary extraction. If there are breaks or missing phenomena in the edge detail data after secondary extraction, the broken parts are repaired through edge connection technology, the continuity of edge details is judged, and the repaired edge detail information is obtained. The edge detection result optimization unit analyzes the edge integrity based on the repaired edge detail information, obtains integrity evaluation indicators, determines whether the edge details meet the preset integrity standards, obtains integrity analysis results, and then adjusts the display parameters of the edge details in combination with the overall features of the detection image to obtain the final edge detection result image. If there are still local blurred areas in the final edge detection result image, local sharpening technology is used to enhance the blurred areas to obtain the optimized final edge detection image.

[0041] Specifically, the multi-scale edge extraction and optimization module significantly improves the completeness and accuracy of edge detection, generating high-quality edge detection result images. This module removes noise interference through image preprocessing, performs multi-scale edge detection based on dynamic scene features, and uses edge operators for secondary extraction to repair broken or missing edge details. Finally, it optimizes the edge detection results through integrity assessment and local sharpening techniques to ensure edge clarity and continuity. This series of methods effectively solves the problem of incomplete and blurry edge detection in complex dynamic environments, providing high-quality edge images for subsequent image analysis and processing, and significantly enhancing the robustness and reliability of the system.

[0042] Furthermore, after obtaining the final edge detection result image, the process also includes: analyzing real-time processing requirements; if the processing time exceeds the preset threshold range, adjusting the filtering parameters and the computational complexity of feature extraction, and determining the optimized processing flow configuration.

[0043] Furthermore, the optimized processing flow configuration is determined, specifically including: Edge detection results are obtained from the input image. The Canny edge detection algorithm is used to generate a first edge image. If the processing time of the first edge image exceeds the preset threshold range, the Gaussian filter parameters are adjusted by reducing the filter kernel size to generate a second edge image. Feature points are extracted from the second edge image, and the Harris corner detection algorithm is used to obtain a feature point set. When the number of feature points exceeds the computational complexity limit, the feature extraction computational complexity is reduced by decreasing the sampling density, and an optimized feature point set is generated. Key regions of the image are obtained from the optimized feature point set, and a segmentation algorithm is used to obtain a segmented region set. The real-time processing flow is adjusted based on the segmented region set to determine the final processing flow configuration.

[0044] Specifically, real-time processing requirements analysis and optimized process configuration can significantly improve the real-time performance and processing efficiency of edge detection systems. Specific methods include reducing the Gaussian filter kernel size, decreasing feature point sampling density, and optimizing the segmentation of key image regions. This ensures that the edge detection system can maintain high detection accuracy while meeting real-time requirements in complex dynamic scenes, significantly enhancing the system's practicality and adaptability in real-world applications.

[0045] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A video stream edge detection system resistant to motion blur, characterized in that: include: The video stream segmentation module obtains the original video data stream from the continuous frame sequence through the video stream data acquisition module, performs preliminary segmentation processing on the high-speed motion features in the dynamic scene, and obtains the segmented video segment sequence. The motion blur analysis module, based on the segmented video segment sequence, uses a motion estimation algorithm to analyze the displacement vectors between adjacent frames, makes a preliminary judgment on the degree of motion blur based on the magnitude of the change in displacement vectors, and determines the distribution of the blur-affected area. The adaptive filtering and deblurring module obtains an adaptive filtering parameter adjustment model based on the distribution of the blur-affected region, extracts the filter kernel parameters that match the degree of motion blur, and obtains a preliminary deblurred frame image after targeted filtering processing. The edge continuity analysis module analyzes the grayscale variation patterns of pixels between frames through preliminary deblurred frame images, and obtains edge continuity features between adjacent frames by combining the inter-frame feature association mapping method to determine potential edge break regions. The edge restoration and enhancement module extracts supplementary information from the continuous features of adjacent frames for potential edge breakage areas, and uses deep feature fusion technology to perform pixel-level restoration of the breakage areas to obtain the restored edge-enhanced image. The illumination correction module analyzes the degree of interference of illumination changes on edge detection based on the restored edge enhancement image, and obtains the illumination-corrected edge image. The multi-scale edge extraction and optimization module takes the edge image after illumination correction and combines it with the complex environmental features in the dynamic scene to obtain a multi-scale edge detection operator for secondary edge extraction, judges whether the edge details are complete, and obtains the final edge detection result image.

2. The anti-motion blur video stream edge detection system according to claim 1, characterized in that: The video stream segmentation processing module includes an inter-frame difference analysis unit, a motion trajectory tracking unit, and an optical flow analysis and secondary segmentation unit; The inter-frame difference analysis unit is used to capture the input continuous frame data in real time, analyze the continuous frames using the inter-frame difference method, determine the changing areas in the dynamic scene, and obtain the preliminary feature areas of high-speed motion. When the motion trajectory tracking unit detects a preliminary feature region of high-speed motion, it tracks the motion trajectory of that region to determine the continuity characteristics of the moving object. Then, it combines a preset threshold to segment the video data stream on the time axis to obtain preliminary segmented video clips. The optical flow analysis and secondary segmentation unit uses optical flow analysis to extract the high-speed motion characteristics of each segment in the initial segmentation, and obtains the motion intensity distribution of each segment. When the motion intensity of a segment exceeds a preset threshold, the segment is further segmented to determine the final video segment sequence.

3. The anti-motion blur video stream edge detection system according to claim 1, characterized in that: The motion fuzzy analysis module includes a dense optical flow calculation unit, a fuzzy region extraction unit, a fuzziness degree classification unit, and a fuzzy region spatial description unit; The dense optical flow calculation unit acquires adjacent frame pairs, uses the dense optical flow algorithm to calculate the displacement vector field of each pair of adjacent frames, obtains the inter-frame pixel-level displacement vector, calculates the amplitude of each pixel displacement vector, and generates an amplitude distribution map. The fuzzy region extraction unit determines that there is significant motion in a region when the amplitude of a certain region in the amplitude distribution map exceeds a preset threshold. It then obtains the coordinates of the high-amplitude region to obtain the potential fuzzy region and uses Gaussian filtering to smooth the displacement vector field to generate a smoothed vector field. The fuzziness level classification unit determines the boundary distribution of the fuzzy region, extracts the amplitude change trend from the smoothed vector field, and uses the K-means clustering algorithm to classify the amplitude change trend to obtain the fuzziness level distribution. The fuzzy region spatial description unit generates a spatial mask for the fuzzy region based on the distribution of the degree of fuzziness, determines the final distribution of the fuzzy influence region, performs grid division, generates statistical features of the region distribution, and obtains a spatial quantitative description of the fuzzy region.

4. The anti-motion blur video stream edge detection system according to claim 1, characterized in that: The adaptive filtering and deblurring module includes a blurred region analysis unit, a filter parameter matching unit, an adaptive filtering and optimization unit, and a local detail restoration unit; The fuzzy region analysis unit analyzes the fuzzy effects and regional distribution characteristics of the input image to obtain a distribution map of the fuzzy region, determine the range of fuzzy effects, and then, in combination with the relevant characteristics of motion fuzziness and fuzziness degree, judge the distribution pattern of the fuzzy type. When the filter parameter matching unit detects that motion blur occupies the main area, it uses a preset motion trajectory analysis tool to extract the motion direction and intensity, and then uses a pre-established adjustment model to match the filter kernel value corresponding to the degree of blur, and determines the appropriate set of filter kernel parameters. The adaptive filtering and optimization unit performs adaptive filtering on the blurred region in accordance with business requirements, generates an initial frame image with preliminary deblurring, analyzes the residual blurred region after image processing, and performs secondary processing by iteratively adjusting the filter kernel value when the residual blurred region exceeds a preset threshold, to obtain the optimized deblurred frame image. The local detail restoration unit acquires the overall sharpness distribution of the image processing, determines whether there is local blurring, determines the final sharp frame image, and then verifies it using an image quality assessment tool in conjunction with the overall deblurring effect to obtain an output image that meets business requirements.

5. The anti-motion blur video stream edge detection system according to claim 1, characterized in that: The edge continuity analysis module includes an image enhancement and grayscale analysis unit, an inter-frame feature correlation extraction unit, and a break region localization unit; The image enhancement and grayscale analysis unit performs preliminary processing on the blurred frame image, uses image enhancement technology for noise reduction and contrast adjustment, and then scans the pixel grayscale information frame by frame to obtain the grayscale change pattern between frames and determine the areas with significant grayscale changes. The inter-frame feature association extraction unit extracts relevant features between adjacent frames by combining inter-frame feature mapping methods for regions with significant grayscale changes. When there are significantly discontinuous feature values ​​in the inter-frame feature association matrix, edge continuity analysis is performed on the corresponding region to determine the possibility of potential edge breaks. The fracture region localization unit, based on the edge continuity analysis results, uses an edge detection algorithm such as the Canny algorithm to locate potential fracture regions, obtains a preliminary distribution map of the fracture regions, and combines the contextual features between adjacent frames to obtain the precise boundary of the fracture regions and determine the final fracture region range.

6. The anti-motion blur video stream edge detection system according to claim 1, characterized in that: The edge restoration and enhancement module includes a feature extraction and region localization unit, a feature fusion and pixel restoration unit, and an edge enhancement and image fusion unit; The feature extraction and region localization unit obtains inter-frame information from adjacent frame groups, analyzes continuous features using a preset feature extraction method, locates potential regions, and uses image processing technology to mark edge breakage positions to determine the boundary range of the breakage region. The feature fusion and pixel repair unit, when the boundary range of the labeled fracture region is significantly different from the continuous features of the adjacent frame group, integrates the inter-frame information and fracture region features through a deep fusion method, and then uses a convolutional neural network to perform pixel-level processing on the fused feature set to perform pixel repair operation on the edge fracture region. The edge enhancement and image fusion unit uses edge enhancement technology to optimize the boundary of the fractured area, obtain enhanced edge detail information, and performs overall image processing to fuse the global features of the repaired image and generate the final edge-enhanced repaired image.

7. The anti-motion blur video stream edge detection system according to claim 1, characterized in that: The illumination correction module includes an illumination distribution feature extraction unit, a local contrast adjustment unit, and an illumination normalization and optimization unit. The illumination distribution feature extraction unit extracts illumination distribution features from the edge enhancement image, calculates illumination change values ​​using a histogram equalization method, and adjusts the edge enhancement image using a local contrast enhancement algorithm when the illumination change data exceeds a preset threshold range to obtain a contrast-corrected image. The local contrast adjustment unit extracts edge features from the contrast-corrected image, uses the Canny edge detection algorithm, and performs smoothing processing using Gaussian filtering based on the noise distribution in the edge image to obtain a smoothed edge image.

8. The anti-motion blur video stream edge detection system according to claim 7, characterized in that: The illumination normalization and optimization unit extracts illumination correction parameters from the smooth edge image, performs illumination normalization processing using an adaptive histogram equalization method, optimizes edge continuity using morphological processing methods based on the edge intensity of the illumination correction image, extracts the final edge features, and calculates the edge gradient using the Sobel operator to obtain the final edge image.

9. The anti-motion blur video stream edge detection system according to claim 1, characterized in that: The multi-scale edge extraction and optimization module includes an image preprocessing unit, a multi-scale edge detection unit, and an edge detection result optimization unit. The image preprocessing unit preprocesses the edge image after illumination correction and removes noise interference using image smoothing technology to obtain preliminary processed edge image data. The multi-scale edge detection unit acquires environmental feature information in dynamic scenes, applies multi-scale detection methods to extract edge features at different scales, and uses edge operators to perform secondary extraction operations to obtain more refined edge line information, thus obtaining edge detail data after secondary extraction. If there are breaks or missing phenomena in the edge detail data after secondary extraction, the broken parts are repaired through edge connection technology to obtain repaired edge detail information. The edge detection result optimization unit analyzes the edge integrity, obtains integrity evaluation indicators, determines whether the edge details meet the preset integrity standards, and then, in combination with the overall features of the detected image, adjusts the display parameters of the edge details to obtain the final edge detection result image. If there are still local blurred areas in the final edge detection result image, local sharpening technology is used to enhance the blurred areas.

10. The anti-motion blur video stream edge detection system according to claim 9, characterized in that: After obtaining the final edge detection result image, the process also includes: analyzing real-time processing requirements; if the processing time exceeds the preset threshold range, adjusting the filtering parameters and the computational complexity of feature extraction, and determining the optimized processing flow configuration.

Images