AI-based expressway video quality diagnosis method, system, equipment and medium

Abstract

The invention discloses an AI-based highway video quality diagnosis method, system and device and a medium, and relates to the technical field of video quality diagnos.The method comprises the steps that an initial diagnosis parameter set is determined based on obtained monitoring device parameters, environment condition information and quality standard information, and video quality problem points are recognized; analyzing different scene complexity regions of the preprocessed target video stream by using the initial diagnosis parameter set, diagnosing video quality problem points, and obtaining image instruction evaluation indexes and optimization suggestion parameters of each region and the problem points; learning and optimizing a preset imaging characteristic model and a common problem mode library on the basis of the collected historical video samples, determining a target video quality improvement scheme and dynamic adaptive adjustment parameters in combination with image quality evaluation indexes of different scene complexity regions, and optimizing and repairing the video quality; according to the invention, the quality of expressway videos is improved, and the performance and reliability of a video monitoring system are improved.

Description

Technical Field

[0001] The present application relates to the technical field of video quality diagnosis, and in particular to an AI-based highway video quality diagnosis method, system, device and medium. Background Art

[0002] With the rapid development of highway networks, traffic monitoring systems play a vital role in ensuring driving safety, improving management efficiency and facilitating post-traffic incident tracing. However, due to various factors, the video quality of highway video monitoring systems often suffers from camera signal loss, video image freezing, video image blur, video image contrast difference, video image severe shaking, camera deviation from preset position and other video quality problems.

[0003] Existing highway video quality diagnosis includes manual identification by professionals who regularly review video data and automatic detection algorithms that intelligently judge whether the video quality meets the standards based on predefined quality judgment parameters (such as picture brightness threshold, contrast threshold, etc.). However, current technology focuses more on repairing problems after they occur, and lacks preventive maintenance measures. It is difficult to fundamentally solve the problem by mainly processing after the event, and there is a certain lag. Therefore, there is an urgent need for a more intelligent and efficient video quality diagnosis method to significantly improve the performance and reliability of highway video surveillance systems. Summary of the invention

[0004] In order to improve the quality of highway videos and enhance the performance and reliability of video surveillance systems, the present application provides an AI-based highway video quality diagnosis method, system, device and medium.

[0005] In the first aspect, the invention objective of the present application is achieved by adopting the following technical solutions: AI-based highway video quality diagnosis method, including: Preprocessing the target highway monitoring video stream to obtain a preprocessed target video stream; Based on the acquired monitoring equipment parameters, environmental condition information, and quality standard information of each key link in the video acquisition process, determine the initial diagnostic parameter set and identify possible video quality problem points; Analyze different scene complexity areas of the preprocessed target video stream using the initial diagnostic parameter set, and perform focused diagnosis on the identified video quality problem points to obtain image instruction evaluation indicators and corresponding optimization recommendation parameters for each area and problem point; Collect historical video samples according to the monitoring device parameters, learn and optimize the preset imaging characteristic model and common problem pattern library corresponding to the device parameters based on the historical video samples, and determine the target video quality improvement plan and dynamic adaptive adjustment parameters in combination with image quality evaluation indicators in different scene complexity areas; According to the target video quality improvement scheme and the dynamic adaptive adjustment parameters, the monitoring system automatically corrects the instructions and optimizes and repairs the video quality.

[0006] By adopting the above technical solution, the preset imaging characteristic model is associated with AI technology, which is used to identify possible video quality problems using computer vision technology, and can also be used to calculate and determine the target video quality improvement plan and dynamically adaptively adjust parameters; the scene complexity area refers to the different areas in the video image divided according to the complexity of the content and the difficulty of processing, and the video quality problem points refer to the various defects or problems that appear in the monitoring video and affect the image clarity, readability and overall quality, which need to be evaluated and analyzed based on the video frames obtained after frame processing of the video stream; specifically, in order to improve the quality of highway videos and improve monitoring efficiency, the present application first pre-processes the monitoring video, including removing noise, enhancing contrast, optimizing color reproduction, etc., to obtain a pre-processed target video stream with better clarity; then accurately identifies the possible video quality problem points in the highway video, so as to facilitate the subsequent key quality analysis, diagnosis and repair of the video quality problem points; then according to the non The optimal image quality evaluation index and optimization recommendation parameters are automatically generated based on the characteristics of the same scene, so as to enhance the adaptability of monitoring equipment to different scenes on the highway during the video acquisition process of the highway. Furthermore, the present application establishes an imaging characteristic model and a common problem pattern library to predict and prevent problems that may reduce video quality in the future, such as camera signal loss, camera screen occlusion, video screen freezing, video screen blur, video screen contrast difference, video screen severe shaking, camera deviation from preset position and other video quality problems, and then specifically specifies the target video quality improvement plan and dynamic adaptive adjustment parameters to automatically adjust the parameter settings according to the real-time changing environmental conditions and video content, which is conducive to ensuring that the video quality remains in the optimal state in various complex environments. In the future, AI technology will continue to be applied to automatically correct and optimize the quality problems in the monitoring system in real time, so as to improve the traffic management efficiency of highway monitoring, and at the same time improve the performance and reliability of the video surveillance system.

[0007] In a preferred example of the present application, the historical video samples are collected according to the monitoring device parameters, and the preset imaging characteristic model and common problem pattern library corresponding to the device parameters are learned and optimized based on the historical video samples, and the target video quality improvement plan and dynamic adaptive adjustment parameters are determined in combination with the image quality evaluation indicators of different scene complexity areas, specifically including: Acquire historical video samples according to the monitoring device parameters, wherein the historical video samples satisfy the environmental condition information corresponding to the monitoring device parameters and the quality standard information of each key link in the video acquisition process; Determine imaging characteristic information and several common problem patterns based on the historical video samples, and optimize the preset imaging characteristic model corresponding to the monitoring device parameters in combination with the device output stability and image processing performance of the monitoring device parameters; Based on the preset imaging characteristic model and common problem pattern library, different scene complexity areas are analyzed to obtain corresponding imaging characteristics and problem pattern recognition results; Based on the imaging features and problem pattern recognition results, combined with image quality assessment indicators in different scene complexity areas, several target optimization control parameters are determined in the initial diagnostic parameter set; Based on the imaging characteristic model, the common problem pattern library, a number of the target optimization control parameters and the corresponding optimization suggestion parameters, a target video quality development plan and dynamic adaptive adjustment parameters are determined.

[0008] By adopting the above-mentioned technical solution, historical video samples that meet specific environmental conditions and quality standards are obtained based on the monitoring equipment parameters, so as to facilitate learning through AI technology and improve the adaptability and accuracy of the model for highway video quality diagnosis; this application integrates imaging characteristic models, common problem pattern libraries and optimization recommendation parameters to formulate a comprehensive target video quality improvement plan and dynamic adaptive adjustment parameters, which is conducive to improving the recognition accuracy and optimization efficiency of video quality problem points, ensuring the consistency and stability of video quality under different environmental conditions, and enhancing the reliability and practicality of the system.

[0009] In a preferred example of the present application, the dynamic adaptive adjustment parameters include optimization thresholds set based on different environmental conditions and video quality problem points; after triggering the automatic correction instruction of the monitoring system according to the target video quality improvement plan and the dynamic adaptive adjustment parameters, the method includes: Based on the target optimization control parameters, generating a plurality of optimization control units in the different scene complexity areas; In the preset dynamic adjustment output model, each optimization control unit is used as a basic unit, and target optimization values ​​are set for all target optimization control parameters in the optimization control unit, and the total optimization value of all target optimization control parameters is set equal to the preset optimization value sum threshold; Based on the identified current environmental conditions and current video quality issues, a corresponding optimization threshold is obtained; with each optimization control unit as a basic unit, the optimization adjustment values ​​of all target optimization control parameters of each optimization control unit are controlled to be less than or equal to the corresponding optimization threshold.

[0010] By adopting the above technical scheme, multiple optimization control units are conducive to realizing regionalized management and improving the precision of optimization. At the same time, corresponding optimization thresholds are set based on different environmental conditions and different video quality problems, so as to facilitate the subsequent precise optimization through dynamic adaptive adjustment of parameters. In order to improve the efficiency of highway monitoring and realize real-time, targeted and adaptive precise adjustment of video quality of highway monitoring system, this application adopts the traversal method to set the target optimization value in each optimization control unit to ensure that the sum of the overall optimization values ​​meets the preset threshold and maintains the balance of optimization. Based on the current environmental conditions and video quality problem points, the optimization adjustment value of each optimization control unit is controlled not to exceed the corresponding optimization threshold to prevent excessive adjustment, thereby improving the overall performance and reliability of the video monitoring system, which is particularly suitable for complex and changeable environments.

[0011] In a preferred example of the present application, the preset optimization value sum threshold includes a preset first direction optimization total value and a preset second direction optimization total value; the step of setting target optimization values ​​for all target optimization control parameters in the optimization control unit, and setting the optimization value sum of all target optimization control parameters equal to the preset optimization value sum threshold, includes: Using the ergodic method to set target optimization values ​​for all target optimization control parameters in the optimization control unit; In a preset dynamic adjustment output model, a first preset fixed number of multiple target optimization control parameters in the optimization control unit are set, and the sum of the target optimization values ​​in the first direction is equal to the preset first direction optimization total value; A second preset fixed number of multiple target optimization control parameters are set in the optimization control unit, and the sum of the target optimization values ​​in the second direction is equal to the preset second direction optimization total value.

[0012] By adopting the above technical solution, in actual application, the first direction can be set to a horizontal direction, the second direction can be set to a vertical direction, or the first direction can be set to a direction parallel to the length direction of the highway, and the second direction can be set to a height direction perpendicular to the first direction; it can be customized according to the monitoring content and monitoring requirements of the actual highway monitoring equipment; specifically, the present invention realizes the precise allocation of optimization values ​​through balanced distribution of optimization values ​​in the first direction and the second direction, ensures the optimization consistency in different directions, and improves the uniformity and stability of the overall video quality.

[0013] In a preferred example of the present application: the method comprises: setting a benchmark point and a plurality of reference points in a one-to-one correspondence in different scene complexity regions of the preprocessed target video stream, respectively, to obtain a benchmark reference image and corresponding theoretical image data based on the same scene complexity region; After triggering the automatic correction instruction of the monitoring system according to the target video quality improvement scheme and the dynamic adaptive adjustment parameters, the method further includes: Based on the acquired optimized and repaired video frame images, fiducial point images and reference point images of different scene complexity areas are acquired; Based on the benchmark point image and the reference point image in the same scene complexity area, the image quality assessment result of the video frame image is calculated with the corresponding benchmark reference image and theoretical image data as a benchmark.

[0014] By adopting the above technical solution, after optimizing and repairing the video frame image quality, a closed-loop optimization evaluation feedback mechanism is further introduced to facilitate continuous optimization of video quality through AI technology, and each time the video frame image is optimized, targeted adjustments can be made based on the actual results, which is conducive to adapting to complex and changeable environmental conditions and / or maintaining high-efficiency video optimization and repair in mobile scenarios of mobile monitoring systems.

[0015] In a preferred example of the present application: the method for setting the benchmark point and the reference point includes: In the preprocessed target video stream, based on the divided multiple scene complexity regions, in each scene complexity region, a position with set significant features is selected as a reference point, and each reference point corresponds to an initial high-quality image data; A number of reference points are set around the reference point, each of which corresponds to initial theoretical image data under ideal conditions; The initial high-quality image data of all fiducial points and the initial theoretical image data of the reference points are combined to form a fiducial reference image and corresponding theoretical image data.

[0016] By adopting the above technical scheme, the method for setting the benchmark points and reference points ensures a high-quality evaluation basis, and selects positions with significant features in each scene complexity area as benchmark points to ensure the accuracy and representativeness of the evaluation; a number of reference points are set around the benchmark points, corresponding to the initial theoretical image data under ideal conditions, to provide a reliable comparison standard, and the data of all benchmark points and reference points are combined to form a benchmark reference image and corresponding theoretical image data for subsequent evaluation; that is, the selection of the benchmark points and reference points of the present invention is scientific and reasonable, and provides a high-quality evaluation basis.

[0017] In the second aspect, the invention objective of the present application is achieved by adopting the following technical solutions: The AI-based highway video quality diagnosis system is applied to the AI-based highway video quality diagnosis method as described above, and the system includes: A preprocessing module, used for preprocessing the target highway monitoring video stream to obtain a preprocessed target video stream; The initial diagnosis module is used to determine the initial diagnosis parameter set and identify possible video quality problem points based on the acquired monitoring equipment parameters, environmental condition information and quality standard information of each key link in the video acquisition process; An analysis and diagnosis module is used to analyze different scene complexity areas of the preprocessed target video stream using the initial diagnosis parameter set, and to perform focused diagnosis on the identified video quality problem points to obtain image instruction evaluation indicators and corresponding optimization recommendation parameters for each area and problem point; A model optimization module is used to collect historical video samples according to the monitoring device parameters, learn and optimize the preset imaging characteristic model and common problem pattern library corresponding to the device parameters based on the historical video samples, and determine the target video quality improvement plan and dynamic adaptive adjustment parameters in combination with image quality evaluation indicators in different scene complexity areas; The automatic correction module is used to trigger the automatic correction instruction of the monitoring system according to the target video quality improvement plan and the dynamic adaptive adjustment parameters to optimize and repair the video quality.

[0018] In the third aspect, the invention objective of the present application is achieved by adopting the following technical solutions: A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned AI-based highway video quality diagnosis method are implemented.

[0019] In a fourth aspect, the invention objective of the present application is achieved by adopting the following technical solutions: A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned AI-based highway video quality diagnosis method.

[0020] In summary, the present application includes at least one of the following beneficial technical effects: 1. By introducing artificial intelligence technology and dynamic adaptive adjustment strategies, intelligent evaluation and optimization of highway monitoring video quality are achieved. Special emphasis is placed on the identification and targeted resolution of video quality issues, which greatly improves traffic management efficiency and safety, and provides more reliable and efficient visual information support for traffic monitoring; 2. Based on the acquired monitoring equipment parameters, environmental conditions information and key quality standard information in the video acquisition process, the initial diagnostic parameter set is determined and possible video quality problems are identified. By comprehensively considering multiple factors, potential problem points in the video can be more comprehensively identified, the accuracy of problem detection is improved, and the pertinence and effectiveness of diagnosis are ensured; 3. Collect historical video samples and learn to establish imaging characteristic models and common problem pattern libraries that match the current monitoring equipment parameters; through learning from historical data, the system can predict and prevent problems that may arise in the future, take optimization measures in advance, improve the system's foresight and adaptability, and reduce the need for post-processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 is a flow chart of a highway video quality diagnosis method based on AI in one embodiment of the present application; Figure 2 is a flowchart of step S4 in the AI-based highway video quality diagnosis method in one embodiment of the present application; Figure 3 It is a schematic diagram of a device in an embodiment of the present application. DETAILED DESCRIPTION

[0022] The present application is further described in detail below in conjunction with the accompanying drawings.

[0023] In one embodiment, if Figure 1 As shown, the present application discloses an AI-based highway video quality diagnosis method, which specifically includes the following steps: S1: Preprocess the target highway monitoring video stream to obtain a preprocessed target video stream.

[0024] In this embodiment, the target highway monitoring video stream refers to the original video data obtained from the highway monitoring equipment; the preprocessing includes operations such as denoising, contrast enhancement, and color correction, and a deep learning model is used to remove random noise in the highway monitoring video.

[0025] S2: Based on the acquired monitoring equipment parameters, environmental condition information, and quality standard information of each key link in the video acquisition process, determine the initial diagnostic parameter set and identify possible video quality problem points.

[0026] In this embodiment, monitoring equipment parameters include hardware parameters such as camera model, resolution, frame rate, etc.; environmental condition information refers to factors that affect video quality, such as light intensity and weather conditions; quality standard information refers to standards that should be met in each link of the video acquisition process, such as brightness, contrast, etc.; the initial diagnostic parameter set is a set of parameters used to preliminarily evaluate video quality; video quality problem points refer to problem areas that may appear in the video, such as blur, noise, occlusion, etc.

[0027] Specifically, AI-based machine learning algorithms (such as convolutional neural networks) are used to identify quality issues in videos.

[0028] S3: Use the initial diagnostic parameter set to analyze the different scene complexity areas of the preprocessed target video stream, and perform key diagnosis on the identified video quality problem points to obtain image instruction evaluation indicators and corresponding optimization recommendation parameters for each area and problem point.

[0029] In this embodiment, the scene complexity area is a video area divided according to factors such as motion complexity, lighting changes, and whether a traffic accident occurs; the image instruction evaluation index is a specific parameter used to evaluate video quality, such as brightness, contrast, color reproduction, etc.; the optimization suggestion parameter is an improvement measure parameter proposed for the identified problem points.

[0030] Specifically, the video screen is subdivided into multiple areas according to the complexity of the scene, and the identified video quality problems are deeply analyzed to obtain detailed image command evaluation indicators. Based on the evaluation results, optimization parameters are proposed, such as brightness adjustment range, noise reduction degree, or callback position when the camera deviates from the preset position.

[0031] S4: Collect historical video samples according to monitoring equipment parameters, learn and optimize the preset imaging characteristic model and common problem pattern library of the corresponding equipment parameters based on historical video samples, and combine the image quality evaluation indicators of different scene complexity areas to determine the target video quality improvement plan and dynamic adaptive adjustment parameters.

[0032] In this embodiment, historical video samples refer to video data stored in the past, including videos under various environmental conditions. The imaging characteristic model is a mathematical model that describes the imaging characteristics of the device under specific conditions. The common problem pattern library records common video quality problems and their solutions. Dynamic adaptive adjustment parameters refer to parameters that are automatically adjusted according to the real-time environment and video content.

[0033] In this embodiment, Figure 2 As shown, step S4 specifically includes: S41: Acquire historical video samples according to monitoring equipment parameters, where the historical video samples satisfy environmental condition information corresponding to the monitoring equipment parameters and quality standard information of each key link in the video acquisition process.

[0034] Specifically, monitoring equipment parameters include hardware parameters such as camera model, resolution, and frame rate; environmental condition information includes factors that affect video quality, such as light intensity and weather conditions. Quality standard information refers to the standards that should be achieved in each link of the video acquisition process, such as brightness and contrast.

[0035] S42: Determine imaging characteristic information and several common problem patterns based on historical video samples, and optimize a preset imaging characteristic model of corresponding monitoring device parameters in combination with device output stability and image processing performance of the monitoring device parameters.

[0036] Specifically, imaging characteristic information refers to the data on the imaging characteristics of the monitoring equipment under specific conditions, such as color reproduction, contrast, etc.; the common problem mode records common video quality problems; the device output stability gives the performance consistency of the device under different environmental conditions; image processing performance refers to the speed and quality of image processing by the device; in actual application, it is necessary to combine the device output stability and image processing performance, adjust and optimize the preset imaging characteristic model, and confirm through verification testing whether the optimized model can effectively improve the video quality.

[0037] S43: Based on the preset imaging characteristic model and common problem pattern library, analyze the complexity areas of different scenes to obtain corresponding imaging characteristics and problem pattern recognition results.

[0038] Specifically, the preset imaging characteristic model is a mathematical model of imaging characteristics after learning and optimization, and the common problem pattern library is a database for storing common video quality problems and their solutions; the problem pattern recognition result includes the identified video problem pattern and its severity, where the severity is divided into severity levels and represented by different numbers.

[0039] S44: Based on the imaging features and problem pattern recognition results, combined with image quality assessment indicators in different scene complexity areas, several target optimization control parameters are determined in the initial diagnosis parameter set.

[0040] Specifically, the target optimization control parameters are improvement measure parameters finally proposed for the identified video quality problem points.

[0041] S45: Determine a target video quality development plan and dynamic adaptive adjustment parameters based on an imaging characteristic model, a common problem pattern library, a number of target optimization control parameters and corresponding optimization recommendation parameters.

[0042] Specifically, the target video quality development plan refers to a video quality optimization plan within a specified time period determined based on the imaging characteristics of the current monitoring equipment under current environmental conditions; dynamic intelligent adaptive adjustment parameters refer to parameters that are automatically adjusted according to the real-time environment and video content based on a pre-set detection frequency within the same time period.

[0043] S5: According to the target video quality improvement plan and the dynamic adaptive adjustment parameters, the monitoring system is triggered to automatically correct the instructions and optimize and repair the video quality.

[0044] In this embodiment, the highway monitoring system repairs the video quality problem points according to the received monitoring system automatic correction instructions; the dynamic adaptive adjustment parameters include optimization thresholds set based on different environmental conditions and corresponding video quality problem points.

[0045] In step S5 of this embodiment, it specifically includes: S51: Based on a plurality of target optimization control parameters, a plurality of optimization control units are generated in different scene complexity regions.

[0046] In this embodiment, the optimization control unit refers to a basic unit for refined management and optimization processing.

[0047] Specifically, the video screen is subdivided into multiple scene complexity areas to ensure that each area can receive targeted optimization processing; based on the initial diagnostic parameter set and problem pattern recognition results, a number of target optimization control parameters are determined; in each scene complexity area, a number of optimization control units are generated based on these target optimization control parameters, and each unit contains a set of specific optimization parameters.

[0048] S52: In the preset dynamic adjustment output model, each optimization control unit is taken as a basic unit, and target optimization values ​​are set for all target optimization control parameters in the optimization control unit, and the total optimization value of all target optimization control parameters is set equal to the preset optimization value sum threshold.

[0049] Specifically, the dynamic adjustment output model is a mathematical model pre-set in the monitoring system, which is used to guide the adjustment of the optimization control parameters; the target optimization value refers to the specific value that each optimization control parameter should reach; the optimization value sum threshold refers to the preset condition that the sum of the optimization values ​​of all optimization control parameters should meet.

[0050] Further, step S52 includes: The preset optimization value sum threshold includes a preset first direction optimization total value and a preset second direction optimization total value; setting target optimization values ​​for all target optimization control parameters in the optimization control unit, and setting the optimization value sum of all target optimization control parameters equal to the preset optimization value sum threshold, including: S521: Use the traversal method to set target optimization values ​​for all target optimization control parameters in the optimization control unit.

[0051] Specifically, a preliminary target optimization value is set for each target optimization control parameter one by one using the traversal method, ensuring that all target optimization control parameters are reasonably assigned optimization values ​​and avoiding missing any key parameters.

[0052] S522: In a preset dynamic adjustment output model, a first preset fixed number of multiple target optimization control parameters are set in the optimization control unit, and the sum of the target optimization values ​​in the first direction is equal to the preset first direction optimization total value.

[0053] Specifically, the first direction is the horizontal direction, which can be set along the horizontal direction of the length of the highway in the monitoring screen. The first preset fixed number refers to a pre-set number to ensure that there are enough parameters in this direction to participate in the optimization without causing unnecessary computing losses due to excessive number of parameter optimizations; the total optimization value of the first direction refers to the total optimization value threshold set in the first direction (such as the horizontal direction).

[0054] S523: Setting a second preset fixed number of multiple target optimization control parameters in the optimization control unit, where the sum of the target optimization values ​​in the second direction is equal to the preset second direction optimization total value.

[0055] In this embodiment, the second direction optimization total value refers to the optimization value sum threshold set in another direction (such as the vertical direction); the second preset fixed number is a pre-set number to ensure that there are enough parameters involved in the optimization in this direction.

[0056] Specifically, the second direction is a vertical direction perpendicular to the first direction. In actual application, the first direction and the second direction can be customized according to actual needs; the first preset fixed number and the second preset fixed number need to be reasonably set to ensure that the monitoring video optimization process is efficient and accurate. Too many parameters may cause excessive computing load and affect real-time performance, and too few parameters may cause poor optimization effect. Based on the device performance of the monitoring device (high-performance devices can produce more optimization parameters), scene complexity (for example, the entrance of a highway tunnel requires more parameters to handle rapidly changing lighting and motion blur problems), environmental conditions (for example, in an environment with drastic lighting changes, more parameters may be required to handle changes in brightness and contrast; in a stable environment, the number of parameters can be reduced to improve efficiency) and quality standard requirements (for example, for high-demand application scenarios (such as security monitoring), more parameters may be required to ensure high-quality video output). The first preset fixed number and the second preset fixed number are preliminarily set based on the above information. In actual application, the number of parameters can also be dynamically adjusted according to real-time environmental conditions and video quality problems-for example, the number of brightness and contrast related parameters is increased when the lighting changes drastically, and the number of parameters is reduced in a stable environment to improve processing efficiency.

[0057] Therefore, under different environmental conditions, the first preset fixed number and the second preset fixed number are different. Assume that in a highway tunnel monitoring system, the first preset fixed number and the second preset fixed number are set as follows: Entrance section: The lighting changes dramatically and the motion complexity is high, so a larger number of first preset fixed numbers (such as 10) are set to process brightness and contrast, and a larger number of second preset fixed numbers (such as 8) are set to process motion blur in the vertical direction.

[0058] Middle section: The lighting is relatively stable, but there may be noise problems caused by vehicle exhaust, so a smaller first preset fixed number (such as 6) is set to deal with noise in the horizontal direction, and an appropriate second preset fixed number (such as 7) is set to deal with noise in the vertical direction.

[0059] Exit segment: The lighting changes rapidly, but the motion complexity is low, so a larger first preset fixed number (such as 9) is set to handle brightness changes, and a smaller second preset fixed number (such as 5) is set to handle simple problems in the vertical direction.

[0060] S53: Based on the identified current environmental conditions and current video quality issues, obtain the corresponding optimization threshold; taking each optimization control unit as a basic unit, control the optimization adjustment values ​​of all target optimization control parameters of each optimization control unit to be less than or equal to the corresponding optimization threshold.

[0061] Specifically, the current environmental conditions and video quality problem points are identified in real time; based on the identification results, the corresponding optimization thresholds are obtained from the preset optimization threshold library; with each optimization control unit as the basic unit, it is ensured that the optimization adjustment values ​​of all target optimization control parameters therein do not exceed the corresponding optimization thresholds to prevent new problems caused by excessive adjustment.

[0062] For example, in a busy highway tunnel, video quality is easily reduced due to changes in light intensity, the impact of vehicle exhaust, and changes in temperature and humidity caused by frequent vehicles entering and exiting. In actual applications, multiple cameras are installed to monitor traffic conditions in real time. Due to the complex and changeable environmental conditions in the tunnel and possible video quality issues (such as blurred images, increased noise, reduced contrast, etc.), traditional fixed parameter settings are difficult to ensure the consistency and stability of video quality.

[0063] In step S51, the video stream captured by the camera in the tunnel can be first subdivided into multiple scene complexity areas, such as the entrance segment, the middle segment (which can be further subdivided, and a middle segment is taken as an example here) and the exit segment; in each area, based on the identified problem points (such as strong light interference in the entrance segment, low light and high motion complexity in the middle segment, and rapidly changing light in the exit segment), a number of target optimization control parameters (such as brightness, contrast, noise reduction level, etc.) are set, and by generating a number of optimization control units, it is ensured that each unit contains a set of specific optimization parameters for refined management.

[0064] In step S52, a preset dynamic adjustment output model (such as a neural network model or a linear regression model) is used to set specific target optimization values ​​for all target optimization control parameters in each optimization control unit; ensure that the sum of the optimization values ​​of all target optimization control parameters in each optimization control unit is equal to the preset optimization value sum threshold, for example: for the optimization control unit of the entrance section, set the sum of the optimization values ​​of brightness and contrast to be equal to the preset threshold to maintain a balanced improvement in the overall image quality.

[0065] In step S53, the environmental conditions (such as light intensity, temperature, humidity) and video quality problems (such as blur, noise, low contrast) in the tunnel are monitored in real time. According to the current environmental conditions and problems, the corresponding optimization threshold is obtained from the preset optimization threshold library; for example, when it is detected that the entrance section is too bright, the maximum allowable value of the brightness adjustment is obtained as the optimization threshold.

[0066] Take each optimization control unit as the basic unit to ensure that the optimization adjustment values ​​of all target optimization control parameters within it do not exceed the corresponding optimization thresholds to prevent new problems caused by excessive adjustment; for example, for the brightness adjustment of the entrance section, ensure that the adjusted brightness value does not exceed the preset maximum allowable value.

[0067] In one embodiment, the AI-based highway video quality diagnosis method further includes: S100: setting a benchmark point and a plurality of reference points in a one-to-one correspondence in different scene complexity regions of a preprocessed target video stream, and obtaining a benchmark reference image and corresponding theoretical image data based on the same scene complexity region.

[0068] In this embodiment, the reference point refers to a key position in the target video stream that has significant features and is not easily affected by environmental conditions; the reference point is an auxiliary evaluation point set around the reference point, which is used to calibrate and verify the optimization effect; the reference reference image is a high-quality initial image containing the reference point, the reference point and its surrounding area; the theoretical image data refers to the image feature data of the reference point and its surrounding area under ideal conditions; the reference point and the reference point are based on the image identification points in the target video stream, and a unique reference point is set at the same position and the same monitoring angle of the same monitoring device. The reference point is dynamically set based on the change of the reference point.

[0069] Specifically, high-quality initial image data is acquired for each fiducial point and reference point to form a fiducial reference image, and theoretical image data for each reference point is generated based on image features under ideal conditions.

[0070] After step S5, the AI-based highway video quality diagnosis method further includes: S501: Based on the acquired optimized and repaired video frame images, the fiducial point images and reference point images of different scene complexity areas are acquired.

[0071] In this embodiment, the target video stream is subjected to frame extraction processing to obtain a corresponding video frame image; the video frame image that has completed optimization and repair refers to the video frame of the target video stream after the repair and optimization processing; the reference point image refers to the image containing the reference point after optimization; and the reference point image refers to the image containing the reference point after optimization.

[0072] Specifically, images containing benchmark points and reference points are extracted from the optimized and repaired video frames, and the extracted benchmark point images and reference point images are aligned with the preset benchmark reference images to ensure the consistency of comparison. The benchmark point images and reference point images of each scene complexity area are recorded to provide an evaluation basis for subsequent highway video quality diagnosis.

[0073] Furthermore, the method for setting the datum point and the reference point includes: S10: In the preprocessed target video stream, based on the divided multiple scene complexity areas, in each scene complexity area, a position with set significant features is selected as a reference point, and each reference point corresponds to an initial high-quality image data.

[0074] Specifically, the video screen is subdivided into multiple areas of different scene complexity, and then computer vision algorithms (such as edge detection and feature point detection) are used to identify locations with significant features in each area, such as road signs, lane lines, etc.; several locations with significant features are selected as reference points in each scene complexity area to ensure that these points are evenly distributed and cover the entire area; high-quality initial image data is obtained for each reference point to ensure that these images are collected under ideal conditions to provide high clarity and accurate color reproduction.

[0075] S20: setting a number of reference points around the reference point, each reference point corresponding to initial theoretical image data under ideal conditions.

[0076] Specifically, a number of reference points are set around each reference point to ensure that these points can fully reflect the image characteristics of the area; for example, a reference point can be set above, below, left and right of the reference point.

[0077] Under simulated ideal conditions (such as standard illumination and no noise), generate corresponding initial theoretical image data for each reference point. This can be obtained through laboratory testing or historical high-quality sample data; record the initial theoretical image data of each reference point, including brightness, contrast, color information, etc., to form a complete theoretical image library.

[0078] S30: combining the initial high-quality image data of all benchmark points and the initial theoretical image data of the reference points to form a benchmark reference image and corresponding theoretical image data.

[0079] Specifically, the initial high-quality image data of all benchmark points and the initial theoretical image data of the reference points are integrated to form a complete benchmark reference image and a corresponding theoretical image data set, ensuring the data correlation between each benchmark point and the reference points around it, facilitating subsequent quality assessment and optimization adjustment, and storing the formed benchmark reference image and the corresponding theoretical image data in the system database for easy access and update at any time.

[0080] S502: Based on the benchmark point image and the reference point image in the same scene complexity area, the image quality assessment result of the video frame image is calculated with the corresponding benchmark reference image and theoretical image data as a benchmark.

[0081] In this embodiment, the image quality assessment result includes the compliance of all image quality assessment indicators to determine whether the optimization effect meets expectations. Step S501 ensures the complete extraction and accurate registration of the optimized image, provides reliable data support for quality assessment, and improves the accuracy of the assessment results. Step S502 ensures the objective evaluation of the optimization effect through a scientific and reasonable evaluation method, provides reliable feedback for subsequent parameter adjustments, and improves the adaptability and robustness of the monitoring system.

[0082] It should be understood that the serial numbers of the steps in the above embodiments do not imply a sequence of execution. The execution sequence of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0083] In one embodiment, an AI-based highway video quality diagnosis system is provided, and the AI-based highway video quality diagnosis system corresponds to the AI-based highway video quality diagnosis method in the above-mentioned embodiment.

[0084] The AI-based highway video quality diagnosis system includes a preprocessing module, an initial diagnosis module, an analysis and diagnosis module, a model optimization module, and an automatic correction module. The detailed description of each functional module is as follows: A preprocessing module, used for preprocessing the target highway monitoring video stream to obtain a preprocessed target video stream; The initial diagnosis module is used to determine the initial diagnosis parameter set and identify possible video quality problem points based on the acquired monitoring equipment parameters, environmental condition information and quality standard information of each key link in the video acquisition process; The analysis and diagnosis module is used to analyze the different scene complexity areas of the pre-processed target video stream using the initial diagnosis parameter set, and to perform key diagnosis on the identified video quality problem points to obtain the image instruction evaluation indicators of each area and problem point and the corresponding optimization recommendation parameters; The model optimization module is used to collect historical video samples according to the monitoring equipment parameters, learn and optimize the preset imaging characteristic model and common problem pattern library of the corresponding equipment parameters based on the historical video samples, and determine the target video quality improvement plan and dynamic adaptive adjustment parameters in combination with the image quality evaluation indicators of different scene complexity areas; The automatic correction module is used to trigger the automatic correction instructions of the monitoring system according to the target video quality improvement plan and dynamic adaptive adjustment parameters to optimize and repair the video quality.

[0085] Optionally, the model optimization module includes: A sample collection unit is used to obtain historical video samples according to monitoring equipment parameters, and the historical video samples meet the environmental condition information corresponding to the monitoring equipment parameters and the quality standard information of each key link in the video acquisition process; A model building unit, for determining imaging characteristic information and several common problem patterns based on historical video samples, and optimizing a preset imaging characteristic model corresponding to the monitoring device parameters in combination with device output stability and image processing performance of the monitoring device parameters; A regional analysis unit is used to analyze regions of different scene complexity based on a preset imaging characteristic model and a common problem pattern library to obtain corresponding imaging characteristics and problem pattern recognition results; A parameter determination unit, used to determine a number of target optimization control parameters in an initial diagnostic parameter set based on imaging features and problem pattern recognition results and in combination with image quality assessment indicators of different scene complexity areas; The solution generation unit is used to determine the target video quality improvement solution and dynamic adaptive adjustment parameters based on the imaging characteristic model, the common problem pattern library, several target optimization control parameters and corresponding optimization suggestion parameters.

[0086] For the specific limitations of the AI-based highway video quality diagnosis system, please refer to the limitations of the AI-based highway video quality diagnosis method in the above text, which will not be repeated here; the various modules in the above-mentioned AI-based highway video quality diagnosis system can be implemented in whole or in part through software, hardware and their combination; the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the corresponding operations of the above modules.

[0087] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store video quality problem points, target video quality improvement solutions and dynamic adaptive adjustment parameters, etc. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a highway video quality diagnosis method based on AI is implemented.

[0088] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program: S1: preprocessing the target highway monitoring video stream to obtain a preprocessed target video stream; S2: Based on the acquired monitoring equipment parameters, environmental condition information, and quality standard information of each key link in the video acquisition process, determine the initial diagnostic parameter set and identify possible video quality problem points; S3: Analyze different scene complexity areas of the preprocessed target video stream using the initial diagnostic parameter set, and perform focused diagnosis on the identified video quality problem points to obtain image instruction evaluation indicators for each area and problem point and the corresponding optimization recommendation parameters; S4: Collect historical video samples according to monitoring equipment parameters, learn and optimize the preset imaging characteristic model and common problem pattern library of the corresponding equipment parameters based on historical video samples, and determine the target video quality improvement plan and dynamic adaptive adjustment parameters in combination with image quality evaluation indicators in different scene complexity areas; S5: According to the target video quality improvement plan and the dynamic adaptive adjustment parameters, the monitoring system is triggered to automatically correct the instructions and optimize and repair the video quality.

[0089] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented: S1: preprocessing the target highway monitoring video stream to obtain a preprocessed target video stream; S2: Based on the acquired monitoring equipment parameters, environmental condition information, and quality standard information of each key link in the video acquisition process, determine the initial diagnostic parameter set and identify possible video quality problem points; S3: Analyze different scene complexity areas of the preprocessed target video stream using the initial diagnostic parameter set, and perform focused diagnosis on the identified video quality problem points to obtain image instruction evaluation indicators for each area and problem point and the corresponding optimization recommendation parameters; S4: Collect historical video samples according to monitoring equipment parameters, learn and optimize the preset imaging characteristic model and common problem pattern library of the corresponding equipment parameters based on historical video samples, and determine the target video quality improvement plan and dynamic adaptive adjustment parameters in combination with image quality evaluation indicators in different scene complexity areas; S5: According to the target video quality improvement plan and the dynamic adaptive adjustment parameters, the monitoring system is triggered to automatically correct the instructions and optimize and repair the video quality.

[0090] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0091] Those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0092] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, it should be understood by those skilled in the art that the technical solutions described in the aforementioned embodiments may still be modified, or some of the features thereof may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. The AI-based highway video quality diagnosis method is characterized by: include: Preprocessing the target highway monitoring video stream to obtain a preprocessed target video stream; Based on the acquired monitoring equipment parameters, environmental condition information, and quality standard information of each key link in the video acquisition process, determine the initial diagnostic parameter set and identify possible video quality problem points; Analyze different scene complexity areas of the preprocessed target video stream using the initial diagnostic parameter set, and perform focused diagnosis on the identified video quality problem points to obtain image instruction evaluation indicators and corresponding optimization recommendation parameters for each area and problem point; Collect historical video samples according to the monitoring device parameters, learn and optimize the preset imaging characteristic model and common problem pattern library corresponding to the device parameters based on the historical video samples, and determine the target video quality improvement plan and dynamic adaptive adjustment parameters in combination with image quality evaluation indicators in different scene complexity areas; According to the target video quality improvement scheme and the dynamic adaptive adjustment parameters, the monitoring system automatically corrects the instructions and optimizes and repairs the video quality.

2. The AI-based highway video quality diagnosis method according to claim 1 is characterized in that: The collecting of historical video samples according to the monitoring device parameters, learning and optimizing the preset imaging characteristic model and common problem pattern library corresponding to the device parameters based on the historical video samples, and combining the image quality evaluation indicators of different scene complexity areas to determine the target video quality improvement plan and dynamic adaptive adjustment parameters, specifically includes: Acquire historical video samples according to the monitoring device parameters, wherein the historical video samples satisfy the environmental condition information corresponding to the monitoring device parameters and the quality standard information of each key link in the video acquisition process; Determine imaging characteristic information and several common problem patterns based on the historical video samples, and optimize the preset imaging characteristic model corresponding to the monitoring device parameters in combination with the device output stability and image processing performance of the monitoring device parameters; Based on the preset imaging characteristic model and common problem pattern library, different scene complexity areas are analyzed to obtain corresponding imaging characteristics and problem pattern recognition results; Based on the imaging features and problem pattern recognition results, combined with image quality assessment indicators in different scene complexity areas, several target optimization control parameters are determined in the initial diagnostic parameter set; Based on the imaging characteristic model, the common problem pattern library, a number of the target optimization control parameters and the corresponding optimization suggestion parameters, a target video quality development plan and dynamic adaptive adjustment parameters are determined.

3. The AI-based highway video quality diagnosis method according to claim 2 is characterized in that: The dynamic adaptive adjustment parameters include optimization thresholds set based on different environmental conditions and video quality problem points; After triggering the automatic correction instruction of the monitoring system according to the target video quality improvement scheme and the dynamic adaptive adjustment parameters, the method includes: Based on the target optimization control parameters, generating a plurality of optimization control units in the different scene complexity areas; In the preset dynamic adjustment output model, each optimization control unit is used as a basic unit, and target optimization values ​​are set for all target optimization control parameters in the optimization control unit, and the total optimization value of all target optimization control parameters is set equal to the preset optimization value sum threshold; Based on the identified current environmental conditions and current video quality issues, a corresponding optimization threshold is obtained; with each optimization control unit as a basic unit, the optimization adjustment values ​​of all target optimization control parameters of each optimization control unit are controlled to be less than or equal to the corresponding optimization threshold.

4. The AI-based highway video quality diagnosis method according to claim 3 is characterized in that: The preset optimization value sum threshold includes a preset first direction optimization total value and a preset second direction optimization total value; the step of setting target optimization values ​​for all target optimization control parameters in the optimization control unit and setting the optimization value sum of all target optimization control parameters equal to the preset optimization value sum threshold includes: Using the ergodic method to set target optimization values ​​for all target optimization control parameters in the optimization control unit; In a preset dynamic adjustment output model, a first preset fixed number of multiple target optimization control parameters in the optimization control unit are set, and the sum of the target optimization values ​​in the first direction is equal to the preset first direction optimization total value; A second preset fixed number of multiple target optimization control parameters are set in the optimization control unit, and the sum of the target optimization values ​​in the second direction is equal to the preset second direction optimization total value.

5. The AI-based highway video quality diagnosis method according to claim 3 is characterized in that: The method comprises: setting a benchmark point and a plurality of reference points in a one-to-one correspondence in different scene complexity regions of the preprocessed target video stream, respectively, to obtain a benchmark reference image based on the same scene complexity region and corresponding theoretical image data; After triggering the automatic correction instruction of the monitoring system according to the target video quality improvement scheme and the dynamic adaptive adjustment parameters, the method further includes: Based on the acquired optimized and repaired video frame images, fiducial point images and reference point images of different scene complexity areas are acquired; Based on the benchmark point image and the reference point image in the same scene complexity area, the image quality assessment result of the video frame image is calculated with the corresponding benchmark reference image and theoretical image data as a benchmark.

6. The AI-based highway video quality diagnosis method according to claim 5 is characterized in that: The method for setting the benchmark point and the reference point includes: In the preprocessed target video stream, based on the divided multiple scene complexity regions, in each scene complexity region, a position with set significant features is selected as a reference point, and each reference point corresponds to an initial high-quality image data; A number of reference points are set around the reference point, each of which corresponds to initial theoretical image data under ideal conditions; The initial high-quality image data of all fiducial points and the initial theoretical image data of the reference points are combined to form a fiducial reference image and corresponding theoretical image data.

7. AI-based highway video quality diagnosis system, characterized by: Applied to the AI-based highway video quality diagnosis method according to any one of claims 1 to 6, the system comprises: A preprocessing module, used for preprocessing the target highway monitoring video stream to obtain a preprocessed target video stream; The initial diagnosis module is used to determine the initial diagnosis parameter set and identify possible video quality problem points based on the acquired monitoring equipment parameters, environmental condition information and quality standard information of each key link in the video acquisition process; An analysis and diagnosis module is used to analyze different scene complexity areas of the preprocessed target video stream using the initial diagnosis parameter set, and to perform focused diagnosis on the identified video quality problem points to obtain image instruction evaluation indicators and corresponding optimization recommendation parameters for each area and problem point; A model optimization module is used to collect historical video samples according to the monitoring device parameters, learn and optimize the preset imaging characteristic model and common problem pattern library corresponding to the device parameters based on the historical video samples, and determine the target video quality improvement plan and dynamic adaptive adjustment parameters in combination with image quality evaluation indicators in different scene complexity areas; The automatic correction module is used to trigger the automatic correction instruction of the monitoring system according to the target video quality improvement plan and the dynamic adaptive adjustment parameters to optimize and repair the video quality.

8. The AI-based highway video quality diagnosis system according to claim 7 is characterized in that: The model optimization module includes: A sample collection unit, configured to obtain historical video samples according to the monitoring device parameters, wherein the historical video samples satisfy the environmental condition information corresponding to the monitoring device parameters and the quality standard information of each key link in the video acquisition process; A model building unit, configured to determine imaging characteristic information and a number of common problem patterns based on the historical video samples, and optimize a preset imaging characteristic model corresponding to the monitoring device parameters in combination with device output stability and image processing performance of the monitoring device parameters; A region analysis unit, configured to analyze regions of different scene complexity based on the preset imaging characteristic model and the common problem pattern library, and obtain corresponding imaging characteristics and problem pattern recognition results; A parameter determination unit, configured to determine a plurality of target optimization control parameters in the initial diagnostic parameter set based on the imaging features and the problem pattern recognition results and in combination with image quality assessment indicators of different scene complexity areas; The solution generation unit is used to determine the target video quality improvement solution and the dynamic adaptive adjustment parameters based on the imaging characteristic model, the common problem pattern library, a number of the target optimization control parameters and the corresponding optimization suggestion parameters.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the AI-based highway video quality diagnosis method as described in any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the AI-based highway video quality diagnosis method as described in any one of claims 1 to 6 are implemented.

Images