A city-level monitoring video quality evaluation method and system

Abstract

The application discloses a kind of urban level monitoring video quality evaluation method and system, the method includes: obtaining quality evaluation task;According to quality evaluation task and evaluation processing time, task scheduling is carried out, and task scheduling result is obtained;According to task scheduling result and task important factor, task allocation is carried out, and task allocation result is obtained;Based on task allocation result, the performance of each monitoring camera is evaluated, and the quality evaluation detection result of single monitoring camera of single monitoring camera is obtained;Based on the quality evaluation detection result of single monitoring camera and the score of other regional monitoring video, the overall evaluation of monitoring video in quality evaluation task is carried out, and the urban level monitoring video quality evaluation result is obtained.The application optimally allocates monitoring camera to be processed, so that monitoring video quality evaluation behavior can be quickly carried out, and the speed of urban level monitoring video evaluation can be accelerated.

Description

TECHNICAL FIELD

[0001] The present application relates to a kind of urban level monitoring video quality evaluation method and system, belong to video analysis technical field. BACKGROUND

[0002] Video monitoring quality evaluation is the hot problem of present concern, is currently in the large-scale video monitoring system construction period, but city quality monitoring system often faces the problem that cannot quickly handle large quantities of camera, manager is inconvenient maintenance etc.

[0003] Current video quality evaluation method mainly includes subjective evaluation and objective evaluation, subjective video quality evaluation mainly relies on artificial detection, but when needing to handle large quantities of camera, there will be insufficient human resources, low efficiency, low equipment perfection rate, lack of professional analysis report and so on.Problems. At the same time in traffic intersection and shop etc. Place, the requirement to camera is relatively high, needs to guarantee that the quality of monitoring video is very high, once the picture of one or more cameras in monitoring system appears failure, camera effective monitoring area is too small or target effectiveness is too low, that is, cannot shoot the effective characteristics of target, then it can cause great loss. SUMMARY

[0004] The purpose of the present application is to overcome the deficiencies in the prior art, provide an office building thermal comfort control system and method based on deep reinforcement learning, optimally allocate the camera to be processed, so that the monitoring video quality evaluation behavior can be quickly performed, and the urban level monitoring video evaluation speed can be accelerated. To achieve the above purpose, the technical scheme adopted by the present application is as follows:

[0005] In a first aspect, the present application provides a kind of urban level monitoring video quality evaluation method, comprising:

[0006] Obtain quality evaluation task;

[0007] Task scheduling is carried out according to quality evaluation task and evaluation processing time, and task scheduling result is obtained;

[0008] Task allocation is carried out according to task scheduling result and task importance factor, and task allocation result is obtained;

[0009] Based on task allocation result, the performance of each monitoring camera is evaluated, and the quality evaluation detection result of single monitoring camera of single monitoring camera is obtained;

[0010] Based on the quality evaluation detection result of single monitoring camera and the score of other area monitoring video, the whole monitoring video in quality evaluation task is evaluated, and urban level monitoring video quality evaluation result is obtained.

[0011] With reference to the first aspect, further, the task scheduling comprises:

[0012] respectively setting a task identifier for each acquired quality evaluation task, wherein the task identifier comprises parameter configuration information required in the scheduling process and a special mark;

[0013] prioritizing all to-be-processed monitoring cameras in each task: monitoring cameras that are locally deployed and in a local LAN have the highest priority, and the farther the distance and the more the number of hops, the weaker the priority of the monitoring cameras;

[0014] setting a configuration master task queue and a fast processing queue for each video analysis server, wherein the master task queue comprises task identifiers for normal analysis, and the fast processing queue comprises task identifiers for fast insertion of user-specified batch camera videos in specific application scenarios;

[0015] obtaining the number of videos to be processed by each video analysis server according to the number of to-be-processed monitoring cameras in all tasks of each video analysis server and the number of idle video analysis devices, and calculating a task estimated processing time based on a preset desired processing time of each video;

[0016] evaluating whether the overall task estimated processing time meets the expected requirement, if not, obtaining the number of idle video analysis devices in the nearby area, and again calculating the task estimated processing time until the task estimated processing time meets the expected requirement;

[0017] distributing all to-be-processed monitoring cameras to idle video analysis devices in order of priority to complete task scheduling.

[0018] With reference to the first aspect, further, the task scheduling comprises:

[0019] setting a task priority factor according to the priority of the camera video stream: monitoring cameras that are locally deployed and in a local LAN have the first priority, the rest of the monitoring cameras that are locally deployed have the second priority, and monitoring cameras that are not locally deployed have the third priority;

[0020] sorting tasks according to the task importance factor, calculating the average number of videos to be processed by each video analysis server according to the calculated task estimated processing time, distributing video streams to local devices in order, and when the local device distribution value is equal to the estimated average number of videos to be processed by each video analysis server, distributing the rest of the video streams to non-local devices in order to complete task distribution;

[0021] If there is an urgent task, it is directly pushed into the fast processing queue for processing, and the fast processing queue has the highest priority, which is used to suspend the current processing task and concentrate on processing the urgent task.

[0022] In combination with the first aspect, further, the performance of each monitoring camera is evaluated to obtain a single monitoring camera quality evaluation detection result of the single monitoring camera, which includes:

[0023] The monitoring video of the monitoring camera within a specified time is retrieved, and a quality evaluation algorithm is used to detect common video monitoring fault problems to obtain a monitoring fault evaluation score;

[0024] According to the pre-set weight of the monitoring fault evaluation score, the monitoring fault evaluation scores are weighted and summed to obtain an evaluation score of the monitoring camera. The monitoring camera with an evaluation score lower than a pre-set first threshold is unqualified, and the rest of the monitoring cameras continue to be processed;

[0025] According to the monitoring scene of the monitoring video, it is judged whether the installation position of the monitoring camera is reasonable to obtain a monitoring area evaluation score;

[0026] According to the monitoring video, human-vehicle classification detection, face detection, and license plate detection are performed to determine whether the target can be detected and the detection degree to obtain a target effectiveness detection evaluation score;

[0027] According to the pre-set weight, the obtained monitoring fault evaluation score, monitoring area evaluation score, and target effectiveness detection evaluation score are weighted and summed to obtain a single camera evaluation score, which is the quality evaluation detection result of the single monitoring camera.

[0028] In combination with the first aspect, further, the judgment of whether the installation position of the monitoring camera is reasonable includes:

[0029] Factor one: a scene segmentation algorithm is used to remove invalid monitoring scenes, and the ratio of the area of the remaining images to the area of the entire monitoring region is calculated;

[0030] Factor two: moving target detection is performed on the video, including: the monitoring camera points to the sky and no moving target appears; there is a moving target in the video segment retrieved by the monitoring camera; there is no moving target in the video retrieved by the monitoring camera for a long time; the three different situations will cause differences in the processing time of the video analysis algorithm for the camera;

[0031] In combination with the above two factors, it is judged whether the installation position of the monitoring camera is reasonable. When the ratio calculated by factor one is lower than 50%, the installation position of the monitoring camera is unreasonable. When the calculated ratio is higher than 50%, if factor two appears situation one or situation three, the monitoring camera is not installed reasonably, and if factor two appears situation two, the monitoring camera is installed reasonably.

[0032] With reference to the first aspect, further, the overall evaluation of the monitoring video in the quality evaluation task is performed to obtain the city-level monitoring video quality evaluation result, comprising:

[0033] The comprehensive score of the overall performance index of the monitoring video in the quality evaluation task is obtained by weighted averaging according to the individual camera evaluation scores;

[0034] The comprehensive score of the overall performance index of the city-level monitoring video is obtained by weighted summation and then averaging according to the comprehensive scores of the overall performance indexes of the monitoring videos of the remaining areas and the preset zoning weight, which is the city-level monitoring video quality evaluation result.

[0035] The second aspect, the present application provides a kind of city-level monitoring video quality evaluation system, comprising:

[0036] Client: for obtaining quality evaluation task;

[0037] Center management module: for task scheduling according to quality evaluation task and evaluation processing time, obtains task scheduling result;

[0038] Streaming media forwarding module: for task allocation according to task scheduling result and task important factor, obtains task allocation result;

[0039] Video analysis and image processing module: for the performance of each monitoring camera based on task allocation result, obtains the quality evaluation detection result of single monitoring camera of single monitoring camera;

[0040] System quality evaluation module: for the overall evaluation of the monitoring video in the quality evaluation task based on the quality evaluation detection result of single monitoring camera and the score of other area monitoring video, obtains the city-level monitoring video quality evaluation result.

[0041] With reference to the second aspect, further, the video analysis and image processing module comprises:

[0042] Video quality detection unit: for calling monitoring video in the specified time of monitoring camera, detects common video monitoring fault problem using quality evaluation algorithm, obtains monitoring fault evaluation score;Monitoring fault evaluation score is weighted and summed according to the weight of monitoring fault evaluation score preset, to obtain the evaluation score of monitoring camera;

[0043] Effective monitoring area detection unit: for judging the rationality of the installation position of monitoring camera according to the monitoring scene of monitoring video, obtains monitoring area evaluation score;

[0044] Target effectiveness detection unit: used for performing human-vehicle classification detection, face detection, license plate detection on the monitoring video, judging whether the target can be detected and the detection degree, and obtaining a target effectiveness detection evaluation score;

[0045] Evaluation score merging unit: used for performing weighted summation on the obtained monitoring fault evaluation score, monitoring area evaluation score and target effectiveness detection evaluation score according to a preset weight, and obtaining a single camera evaluation score, which is the quality evaluation detection result of the single monitoring camera.

[0046] In a third aspect, the present application provides a computing device, comprising a processor and a storage medium;

[0047] The storage medium is used for storing instructions;

[0048] The processor is used for operating according to the instructions to perform the steps of the method of the first aspect.

[0049] In a fourth aspect, the present application provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to implement the steps of the method of the first aspect.

[0050] Compared with the prior art, the office building thermal comfort control system and method based on deep reinforcement learning provided by the embodiment of the present application have the following beneficial effects:

[0051] The present application acquires a quality evaluation task, performs task scheduling according to the quality evaluation task and an evaluation processing time, obtains a task scheduling result, performs task allocation according to the task scheduling result and a task importance factor, and obtains a task allocation result. Through task scheduling and task allocation, the present application can optimally allocate the monitoring cameras to be processed;

[0052] Based on the task allocation result, the present application evaluates the performance of each monitoring camera, obtains a quality evaluation detection result of the single monitoring camera of the single monitoring camera, and performs overall evaluation on the monitoring video in the quality evaluation task based on the quality evaluation detection result of the single monitoring camera and the scores of other area monitoring videos, thereby obtaining a city-level monitoring video quality evaluation result. The present application can quickly perform the monitoring video quality evaluation behavior and can accelerate the city-level monitoring video evaluation speed. BRIEF DESCRIPTION OF DRAWINGS

[0053] Figure 1 is a flowchart of a city-level monitoring video quality evaluation method provided by the embodiment one of the present application;

[0054] Figure 2 is a flowchart of task scheduling in a city-level monitoring video quality evaluation method provided by the embodiment one of the present application; is a flowchart of task scheduling in a city-level monitoring video quality evaluation method provided by the embodiment one of the present application;

[0055] Figure 3 is a flowchart of a single monitoring camera quality evaluation detection result of a single monitoring camera obtained in a city-level monitoring video quality evaluation method provided by Embodiment I of the present application;

[0056] Figure 4 is a flowchart of a city-level monitoring video quality evaluation result obtained in a city-level monitoring video quality evaluation method provided by Embodiment I of the present application. DETAILED DESCRIPTION

[0057] The present application will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present application, and cannot be used to limit the protection scope of the present application.

[0058] Embodiment I:

[0059] As shown in Figure 1 , the present embodiment provides a city-level monitoring video quality evaluation method, comprising:

[0060] obtaining a quality evaluation task;

[0061] performing task scheduling according to the quality evaluation task and evaluation processing time to obtain a task scheduling result;

[0062] performing task allocation according to the task scheduling result and a task importance factor to obtain a task allocation result;

[0063] performing performance evaluation on each monitoring camera based on the task allocation result to obtain a single monitoring camera quality evaluation detection result of a single monitoring camera;

[0064] performing overall evaluation on the monitoring video in the quality evaluation task based on the single monitoring camera quality evaluation detection result and scores of other regional monitoring videos to obtain a city-level monitoring video quality evaluation result.

[0065] The specific steps include:

[0066] Step 1: obtaining a quality evaluation task.

[0067] Step 2 as shown in Figure 2 , performing task scheduling according to the quality evaluation task and evaluation processing time to obtain a task scheduling result.

[0068] Step 2.1: setting a task identification for each obtained quality evaluation task, wherein the task identification includes parameter configuration information and special marks required in the scheduling process.

[0069] Step 2.2: Prioritize all to-be-processed monitoring cameras in each task: the monitoring cameras in the local area and in the local LAN have the highest priority, and the priority of the monitoring cameras in the remote area and across the network is weaker.

[0070] Step 2.3: Set a configuration master task queue and a fast processing queue for each video analysis server, wherein the master task queue includes task identifiers for normal analysis, and the fast processing queue includes task identifiers for quickly inserting user-specified batch camera videos in specific application scenarios.

[0071] Step 2.4: According to the number M of to-be-processed monitoring cameras in all tasks of each video analysis server and the number n of idle video analysis devices, obtain the number V of videos that each video analysis server needs to process:

[0072] V = M / n (1)

[0073] Based on the preset desired processing time t of each video, calculate the task estimated processing time T:

[0074] T = V*t (2)

[0075] Specifically, the unit of t is minute.

[0076] Step 2.5: Evaluate whether the overall task estimated processing time meets the expected requirements. If not, obtain the number of idle video analysis devices in the nearby area, return to step 2.4, and continue until the task estimated processing time meets the expected requirements.

[0077] Step 2.6: Distribute all to-be-processed monitoring cameras to idle video analysis devices according to the priority order, and complete task scheduling.

[0078] Step 3 as shown in Figure 2 , according to the task scheduling result and the task importance factor, perform task allocation to obtain a task allocation result.

[0079] Step 3.1: According to the priority of the camera video stream, set the task priority factor: the monitoring cameras in the local area and in the local LAN have the first priority, the remaining monitoring cameras in the local area have the second priority, and the monitoring cameras in the remote area have the third priority.

[0080] Step 3.2: Sort the tasks according to the task importance factor, calculate the average number of videos each video analysis server needs to process based on the estimated processing time of the tasks calculated in step 2.4, and distribute the video streams to the local devices in order. When the local device allocation value is equal to the estimated average number of videos each video analysis server needs to process, the remaining video streams are sequentially assigned to non-local devices, and the task allocation is completed.

[0081] Step 3.3: If there is an urgent task, it is directly pushed into the fast processing queue for processing. The fast processing queue has the highest priority and is used to suspend the current processing task and focus on processing the urgent task.

[0082] Step 4 as shown in Figure 3 , based on the task allocation result, the performance of each monitoring camera is evaluated to obtain the quality evaluation detection result of the single monitoring camera.

[0083] Step 4.1: Retrieve the monitoring video of the monitoring camera within a specified time, and use the quality evaluation algorithm to detect common video monitoring failure problems to obtain the monitoring failure evaluation score I.

[0084] Common video monitoring failure problems include: black screen, video signal loss, image clarity, color cast, gray scale, and video jitter.

[0085] The monitoring failure evaluation score I includes the evaluation scores of the six common video monitoring failure problems.

[0086] Step 4.2: According to the pre-set weight of the monitoring failure evaluation score, the monitoring failure evaluation score obtained in step 4.1 is weighted and summed to obtain the evaluation score G of the monitoring camera. The monitoring camera with an evaluation score G lower than the pre-set first threshold G1 is unqualified, and the remaining monitoring cameras continue to be processed.

[0087] G = (a1g1 + a2g2 + a3g3 + a4g4 + a5g5 + a6g6) / 6 (3)

[0088] In equation (3), g1, g2, g3, g4, g5, g6 are the scores of the six common video monitoring failure problems, and a1... a6 are the weights of the monitoring failure evaluation scores of the six common video monitoring failure problems.

[0089] Step 4.3: Analyze the monitoring scene corresponding to the camera.

[0090] Factor one: Use a scene segmentation algorithm to remove invalid monitoring scenes and calculate the ratio of the area occupied by the remaining images to the entire monitoring area.

[0091] Factor two: moving target detection on the video, including: the monitoring camera points to the sky, no moving target appears; the monitoring camera calls the video segment continuously with moving target; the monitoring camera calls the video continuously without moving target; three different situations will cause the difference of the video analysis algorithm on the camera processing time.

[0092] According to the above two factors, it is judged whether the installation position of the monitoring camera is reasonable. When the ratio calculated by factor one is lower than 50%, the installation position of the monitoring camera is unreasonable; when the ratio calculated is higher than 50%, if factor two appears situation one or situation three, the monitoring camera installation is unreasonable, if factor two appears situation two, the monitoring camera installation is reasonable, and the evaluation score U of the monitoring area is obtained.

[0093] Step 4.4: According to the monitoring video, the human-vehicle classification detection, face detection and license plate detection are carried out to judge whether the target can be detected and the detection degree, and the evaluation score R of the target effectiveness detection is obtained.

[0094] Step 4.5: According to the preset weight, the monitoring fault evaluation score, the monitoring area evaluation score and the target effectiveness detection evaluation score obtained in steps 4.2-4.4 are weighted and summed to obtain the evaluation score E of a single camera:

[0095] E=(α1G+α2U+α3R) / 3 (4)

[0096] In formula (4), α1 is the weight of the monitoring fault evaluation score, α2 is the weight of the monitoring area evaluation score, and α3 is the weight of the target effectiveness detection evaluation score.

[0097] If the video quality is unqualified, the unqualified score is displayed, and the larger the score is, the better the monitoring quality of the monitoring camera is. The evaluation score E of a single camera is the quality evaluation detection result of a single monitoring camera.

[0098] Step 5: Based on the quality evaluation detection result of a single monitoring camera and the scores of other area monitoring videos, the overall performance of the monitoring video in the quality evaluation task is evaluated to obtain the city-level monitoring video quality evaluation result.

[0099] Step 5.1: The evaluation score of a single camera obtained in step 4.5 is weighted and averaged to obtain the comprehensive score Z of the overall performance index of the monitoring video in the quality evaluation task.

[0100] Step 5.2: According to the comprehensive scores A1-A n of the overall performance index of the monitoring video of the remaining areas, the comprehensive score F of the overall performance index of the city-level monitoring video is obtained by weighted summation and average according to the preset zoning weight, that is, the city-level monitoring video quality evaluation result.

[0101] The embodiment can optimally allocate the to-be-processed monitoring camera through task scheduling and task allocation, can efficiently process a large number of cameras, and can reduce the network bandwidth pressure caused by a large number of video streams by preferentially processing local video streams on a local device after sorting the to-be-processed cameras according to geographical positions.

[0102] The embodiment can quickly perform the monitoring video quality evaluation behavior and can accelerate the city-level monitoring video evaluation speed by evaluating the performance of each monitoring camera based on the task allocation result, obtaining the quality evaluation detection result of a single monitoring camera, and performing overall evaluation on the monitoring video in the quality evaluation task based on the quality evaluation detection result of the single monitoring camera and the scores of other regional monitoring videos.

[0103] Embodiment two:

[0104] The embodiment provides a city-level monitoring video quality evaluation system, which comprises:

[0105] The client is configured to obtain a quality evaluation task.

[0106] The center management module is configured to perform task scheduling according to the quality evaluation task and an evaluation processing time, and obtain a task scheduling result.

[0107] The stream media forwarding module is configured to perform task allocation according to the task scheduling result and a task importance factor, and obtain a task allocation result.

[0108] The video analysis and image processing module is configured to evaluate the performance of each monitoring camera based on the task allocation result, and obtain a quality evaluation detection result of a single monitoring camera.

[0109] The system quality evaluation module is configured to perform overall evaluation on the monitoring video in the quality evaluation task based on the quality evaluation detection result of the single monitoring camera and the scores of other regional monitoring videos, and obtain a city-level monitoring video quality evaluation result.

[0110] Specifically, the client outputs the obtained task to the center management module, the center management module performs task scheduling, and outputs the task scheduling result to the stream media forwarding module. The stream media forwarding module performs task allocation according to the received task scheduling result and the task importance factor, and sequentially sends the specified monitoring video to the video analysis and image processing module. The video analysis and image processing module performs quality evaluation detection of a single monitoring camera and outputs the result to the system quality evaluation module, and the system quality evaluation module outputs the city-level monitoring video quality evaluation result.

[0111] The video analysis and image processing module comprises:

[0112] Video quality detection unit: for calling the monitoring video of the monitoring camera in a specified time, using a quality evaluation algorithm to detect common video monitoring fault problems, and obtaining a monitoring fault evaluation score; according to a pre-set weight of the monitoring fault evaluation score, the monitoring fault evaluation score is weighted and summed to obtain an evaluation score of the monitoring camera;

[0113] Effective monitoring area detection unit: for judging the rationality of the installation position of the monitoring camera according to the monitoring scene of the monitoring video, and obtaining a monitoring area evaluation score;

[0114] Target effectiveness detection unit: for performing human-vehicle classification detection, face detection, and license plate detection according to the monitoring video, judging whether the target can be detected and the detection degree, and obtaining a target effectiveness detection evaluation score;

[0115] Evaluation score merging unit: for weighting and summing the obtained monitoring fault evaluation score, monitoring area evaluation score, and target effectiveness detection evaluation score according to a pre-set weight, to obtain a single camera evaluation score, which is the quality evaluation detection result of a single monitoring camera.

[0116] Embodiment three:

[0117] The embodiment of the present application provides a computing device, comprising a processor and a storage medium;

[0118] The storage medium is used for storing instructions;

[0119] The processor is used for operating according to the instructions to perform the steps of the method in the embodiment one.

[0120] Embodiment four:

[0121] The embodiment of the present application further provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to realize the steps of the method in the embodiment one.

[0122] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt a computer program product in the form of one or more computer usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer usable program code.

[0123] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks.

[0124] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks.

[0125] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks.

[0126] The above only is the preferred embodiment of the present application, it should be pointed out that, for those skilled in the technical field, without departing from the technical principles of the present application, can also make a number of improvements and variations, these improvements and variations should be regarded as the protection scope of the present application.

Claims

1. A method for evaluating the quality of urban-level monitoring video, characterized in that, The method comprises the following steps: obtaining a quality evaluation task; scheduling the task according to the quality evaluation task and the evaluation processing time to obtain a task scheduling result; scheduling the task, comprising: setting a task identifier for each obtained quality evaluation task, wherein the task identifier comprises parameter configuration information required in the scheduling process and a special mark; sorting all to-be-processed monitoring cameras in each task according to priority: monitoring cameras located in the local and in the local LAN have the highest priority, and the priority of monitoring cameras located farther away and crossing more times is weaker; setting a configuration main task queue and a fast processing queue for each video analysis server, wherein the main task queue comprises task identifiers for normal analysis, and the fast processing queue comprises task identifiers for quickly inserting user-specified batch camera videos in specific application scenarios; obtaining the number of videos to be processed by each video analysis server according to the number of to-be-processed monitoring cameras in all tasks of each video analysis server and the number of idle video analysis devices, and calculating the task estimated processing time based on the expected processing time of each video; evaluating whether the overall task estimated processing time meets the expected requirement, if not, obtaining the number of idle video analysis devices in the nearby area, and calculating the task estimated processing time again until the task estimated processing time meets the expected requirement; allocating all to-be-processed monitoring cameras to idle video analysis devices according to priority to complete task scheduling; allocating the task according to the task scheduling result and the task importance factor to obtain a task allocation result; the task allocation comprises: setting a task priority factor according to the priority of the camera video stream: monitoring cameras located in the local and in the local LAN have the first priority, the rest of the monitoring cameras located in the local have the second priority, and monitoring cameras located outside the local have the third priority; sorting the tasks according to the task importance factor, calculating the average number of videos to be processed by each video analysis server according to the calculated task estimated processing time, and allocating the video stream to the local device in sequence, when the local device allocation value is equal to the estimated average number of videos to be processed by each video analysis server, the remaining video stream is allocated to non-local devices in sequence to complete task allocation; if there is an urgent task, it is directly pushed into the fast processing queue for processing, and the fast processing queue has the highest priority, which is used to suspend the current processing task and concentrate on processing the urgent task; based on the task allocation result, the performance of each monitoring camera is evaluated to obtain the quality evaluation detection result of a single monitoring camera of a single monitoring camera; based on the quality evaluation detection result of a single monitoring camera and the score of other regional monitoring videos, the overall quality of the monitoring videos in the quality evaluation task is evaluated to obtain a city-level monitoring video quality evaluation result.

2. The urban surveillance video quality assessment method of claim 1, wherein, The performance of each monitoring camera is evaluated to obtain the quality evaluation detection result of a single monitoring camera of a single monitoring camera, comprising: The monitoring video of the monitoring camera in a specified time is called, and a quality evaluation algorithm is used to detect common video monitoring fault problems to obtain a monitoring fault evaluation score; wherein the common video monitoring fault problems include: black screen, video signal loss, image definition, color cast, gray scale and video jitter; According to the pre-set weight of the monitoring fault evaluation score, the monitoring fault evaluation score is weighted and summed to obtain the evaluation score of the monitoring camera, and the monitoring camera whose evaluation score is lower than the pre-set first threshold is unqualified, and the rest of the monitoring cameras continue to be processed; According to the monitoring scene of the monitoring video, it is judged whether the installation position of the monitoring camera is reasonable to obtain a monitoring area evaluation score; According to the monitoring video, human-vehicle classification detection, face detection and license plate detection are performed to determine whether the target can be detected and the detection degree to obtain a target effectiveness detection evaluation score; According to the pre-set weight, the obtained monitoring fault evaluation score, monitoring area evaluation score and target effectiveness detection evaluation score are weighted and summed to obtain a single camera evaluation score, which is the quality evaluation detection result of a single monitoring camera.

3. The urban surveillance video quality assessment method of claim 2, wherein, The judgment of whether the installation position of the monitoring camera is reasonable includes: Factor one: using a scene segmentation algorithm to remove invalid monitoring scenes and calculating the ratio of the area of the remaining images to the area of the entire monitoring region; Factor two: performing moving target detection on the video, including: the monitoring camera points to the sky and no moving target appears; there is a moving target in the video segment retrieved by the monitoring camera; there is no moving target in the video retrieved by the monitoring camera for a long time; the three different situations will cause differences in the processing time of the video analysis algorithm for the camera; Combining the above two factors to judge whether the installation position of the monitoring camera is reasonable, when the ratio calculated by factor one is lower than 50%, the installation position of the monitoring camera is unreasonable; when the ratio calculated is higher than 50%, if factor two appears situation one or situation three, the monitoring camera is not installed reasonably, if factor two appears situation two, the monitoring camera is installed reasonably. 4.The urban-level monitoring video quality evaluation method of claim 1, wherein, The overall evaluation of the monitoring video in the quality evaluation task includes: According to the weighted average of the single camera evaluation score, the comprehensive score of the overall performance index of the monitoring video in the quality evaluation task is obtained; According to the comprehensive score of the overall performance index of the monitoring video of the remaining area, the comprehensive score of the overall performance index of the city-level monitoring video is obtained by weighting and averaging according to the pre-set zoning weight, which is the city-level monitoring video quality evaluation result.

5. A city-level monitoring video quality assessment system, characterized in that, It includes: Client: used to obtain quality evaluation tasks; Central management module: used to schedule tasks according to quality evaluation tasks and evaluation processing time to obtain task scheduling results; Task scheduling includes: Setting task identification for each obtained quality evaluation task, wherein the task identification includes parameter configuration information and special marks required in the scheduling process; Prioritize all to-be-processed monitoring cameras in each task: monitoring cameras in the local network have the highest priority, and the farther the monitoring cameras are and the more times they cross, the weaker the priority is; Set a configuration main task queue and a fast processing queue for each video analysis server, wherein the main task queue includes task identifiers for normal analysis, and the fast processing queue includes task identifiers for quickly inserting user-specified batch camera videos in specific application scenarios; According to the number of to-be-processed monitoring cameras in all tasks of each video analysis server and the number of idle video analysis devices, obtain the number of videos that each video analysis server needs to process, and calculate the task estimated processing time based on the preset desired processing time of each video; Evaluate whether the overall task estimated processing time meets the expected requirement, if not, obtain the number of idle video analysis devices in the nearby area, and calculate the task estimated processing time again until the task estimated processing time meets the expected requirement; Assign all to-be-processed monitoring cameras to idle video analysis devices in priority order to complete task scheduling; task allocation includes: According to the priority of camera video streams, set a task priority factor: monitoring cameras in the local network have the first priority, the rest of the monitoring cameras in the local network have the second priority, and monitoring cameras not in the local network have the third priority; According to the task importance factor, sort the tasks, calculate the average number of videos that each video analysis server needs to process according to the calculated task estimated processing time, and assign video streams to local devices in order, when the local device allocation value is equal to the estimated average number of videos that each video analysis server needs to process, assign the remaining video streams to non-local devices in order to complete task allocation; If there is an urgent task, it is directly pushed into the fast processing queue for processing, and the fast processing queue has the highest priority, which is used to suspend the current processing task and concentrate on processing the urgent task; A streaming media forwarding module is configured to perform task allocation according to the task scheduling result and the task importance factor to obtain a task allocation result; A video analysis and image processing module is configured to evaluate the performance of each monitoring camera based on the task allocation result to obtain a quality evaluation detection result of a single monitoring camera; A system quality evaluation module is configured to perform overall evaluation on monitoring videos in a quality evaluation task based on the quality evaluation detection result of a single monitoring camera and scores of monitoring videos in other areas to obtain a city-level monitoring video quality evaluation result.

6. The urban-level monitoring video quality evaluation system of claim 5, wherein, The video analysis and image processing module includes: The video quality detection unit is configured to call monitoring videos of the monitoring camera in a specified time, detect common video monitoring fault problems by using a quality evaluation algorithm, and obtain monitoring fault evaluation scores; the monitoring fault evaluation scores are weighted and summed according to preset weights of the monitoring fault evaluation scores, and an evaluation score of the monitoring camera is obtained; wherein the common video monitoring fault problems include black screen, video signal loss, image definition, color cast, gray scale and video jitter. The effective monitoring area detection unit is configured to judge the rationality of the installation position of the monitoring camera according to the monitoring scene of the monitoring video, and obtain a monitoring area evaluation score. The target effectiveness detection unit is configured to perform human-vehicle classification detection, face detection and license plate detection according to the monitoring video, judge whether a target can be detected and the detection degree, and obtain a target effectiveness detection evaluation score. The evaluation score merging unit is configured to weight and sum the monitoring fault evaluation score, the monitoring area evaluation score and the target effectiveness detection evaluation score according to preset weights, and obtain a single camera evaluation score, which is a quality evaluation detection result of a single monitoring camera.

7. A computing device, comprising: The device comprises a processor and a storage medium. The storage medium is configured to store instructions. The processor is configured to operate according to the instructions to perform the steps of the method of any one of claims 1-4.

8. A computer readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the steps of the method of any one of claims 1-4.

Images