Video management system based on video surveillance platform
Through the video management system of the video surveillance platform, using video synthesis processing technology, short videos containing characters' images are intercepted and customized identification marks are performed, which solves the problems of redundant video data and low viewing efficiency in the video surveillance system, and realizes monitoring video streamlined and quick query.
Patent Information
- Application Number
- CN202311127283.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-09-04
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2043-09-04
AI Technical Summary
The existing video surveillance technology increases memory usage due to the increased clarity of video images. The video data storage period is cleaned after it exceeds, and contains a large number of invalid background images, which makes it inconvenient to view historical surveillance videos.
The video management system based on the video surveillance platform collects monitoring data through infrared cameras, divides it into time period data and transmits it to the service platform for video synthesis processing, and uses video processing units, information analysis units and video synthesis units to analyze human signal characteristics and intercept them. Short videos containing characters' images and stored in media servers to achieve streamlined videos and custom logo marks.
It has achieved streamlining of surveillance videos, improved user viewing efficiency, and customized identification marking and back viewing extraction units to facilitate users to quickly extract required video clips, solving the problem of inconvenience in viewing historical surveillance videos.
Smart Images

Figure CN117176910B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of video management, and in particular to a video management system based on a video monitoring platform. Background Art
[0002] Video surveillance technology is increasingly being used, and the use of video surveillance technology is ubiquitous;
[0003] However, due to the development of cameras, the existing video surveillance technology has obtained clearer and clearer video images, and the memory occupied by video images has also become larger and larger. Video data is stored for a certain period of time and will be cleared after a certain period of time. At the same time, video data contains a large number of background images that do not contain moving subjects. When users view videos, they gradually query them, resulting in a large amount of invalid surveillance video data being watched repeatedly, which in turn causes inconvenience in viewing historical surveillance videos.
[0004] Therefore, the present invention proposes a video management system based on a video monitoring platform, which solves the problem of inconvenience in viewing monitoring videos. Summary of the invention
[0005] In view of the deficiencies of the prior art, the present invention provides a video management system based on a video monitoring platform, which solves the problems raised in the background technology center.
[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: A video management system based on a video monitoring platform, comprising:
[0007] The monitoring platform is used to collect monitoring data captured by infrared cameras in the monitoring area, and then transmit the monitoring data to the business platform, wherein the monitoring data is divided into a plurality of time period data according to a preset time period, and the time period data is transmitted to the business platform one by one in chronological order, and the time period data includes monitoring videos and infrared information of the same time period; the infrared monitoring information is a human signal group obtained at all time points through human infrared sensing technology, and the human signal group includes human signals obtained at each time point;
[0008] A business platform, used for synthesizing surveillance videos, and then transmitting the surveillance videos before and after the synthesis processing to a media server; and it includes a video processing unit, an information analysis unit, and a video synthesis unit. The synthesis processing method is that the video processing unit obtains all video frames and corresponding human body signals of a single time period video, and sends them to the information analysis unit, and then the information analysis unit imports the video frames and human body signals into a pre-trained analysis model for analysis, and then the video synthesis unit obtains the cut-off time period according to the analysis result of the analysis model, and obtains the corresponding short video according to the cut-off time period, and then synthesizes the short videos to obtain a synthesized video, and then sends the synthesized video and the surveillance video to the media server;
[0009] The media server includes an original video library and a key video library, wherein the original video library is used to store the surveillance video before synthesis processing, and the key video library is used to store the synthesized video after synthesis processing.
[0010] Preferably, the human body signal includes two signal characteristics: a non-human signal and a human signal.
[0011] Preferably, the synthesis treatment is as follows:
[0012] In the first step, the video processing unit obtains a single time period video in real time, and divides the time period video into a number of standard time nodes according to the time trend, and the time difference between each adjacent standard time node is the same, and then obtains the corresponding video frame from the time period video according to the standard time node, and obtains the corresponding human body signal in the infrared monitoring information according to the standard time node;
[0013] Then the video processing unit sends all video frames and all human body signals obtained in the video during this period to the information analysis unit;
[0014] Step 2: The information analysis unit imports the video frames and human body signals at the same standard time node into the pre-trained analysis model one by one in chronological order;
[0015] The analysis model obtains the signal characteristics of the human body signal at the standard time point, and determines the human image contained in the video frame based on the face recognition technology, and then transmits the video frame at the corresponding standard time node to the video synthesis unit based on the determination result;
[0016] Step 3: the video synthesis unit sorts the video frames transmitted by the information analysis unit according to the chronological order and generates a synthesis sequence library;
[0017] Then, the time difference between two adjacent standard time nodes is obtained and recorded as the comparison difference;
[0018] Then, the continuity of two adjacent video frames in the synthetic sequence library is judged according to the contrast difference, and then the video frames with continuity are obtained in the synthetic sequence table according to the continuity judgment result, and the standard time nodes of the first and last video frames are obtained therefrom, and then the two standard time nodes are used as the clipping period;
[0019] Then, the corresponding short video is intercepted according to the position corresponding to the interception time period and the video of the time period;
[0020] Afterwards, within a specified period, all short videos of all time periods are synthesized in chronological order to obtain a synthesized video, and then the synthesized video and the original surveillance video are sent to the media server.
[0021] Preferably, the analysis model determines the image of a person in a video frame in the following manner:
[0022] Select a standard time node;
[0023] If the signal feature acquired by the analysis model is a human signal and the video frame contains a human image, the video frame at the standard time node is transmitted to the video synthesis unit;
[0024] If the signal feature acquired by the analysis model is a human signal and the video frame does not contain a human image, the video frame at the standard time node is transmitted to the video synthesis unit;
[0025] If the signal feature obtained by the analysis model is a non-human signal and the video frame contains a human image, the video frame will not be transmitted to the video synthesis unit;
[0026] If the signal feature acquired by the analysis model is a non-human signal and the video frame does not contain a human image, the video frame will not be transmitted to the video synthesis unit.
[0027] Preferably, the continuity determination method of two adjacent video frames is as follows:
[0028] If the time difference between the standard time nodes corresponding to two adjacent video frames is greater than the contrast difference, then there is no continuity between the two adjacent video frames;
[0029] If the time difference between the standard time nodes corresponding to two adjacent video frames is equal to the contrast difference, then the two adjacent video frames are continuous.
[0030] Preferably, the business platform is also used to perform custom identification marking processing on all short videos in the composite video to obtain a composite video with short video identification markings, and then send the composite video with short video identification markings to the media server.
[0031] Preferably, the business platform further comprises an information self-recording unit, an identification marking unit, and a review extraction unit; the custom identification marking processing method is as follows:
[0032] Step 1: Select a group of short videos, the user uploads the user image through the information self-recording unit, and customizes the user image, and transmits the customized named user image to the pre-trained analysis model; wherein the user image is the user's face image;
[0033] Step 2: The identification and marking unit performs face recognition on all face images in the short video through a pre-trained analysis model, obtains the user image corresponding to the face image in the short video based on the user image after custom naming, and identifies the short video, and then sends the synthesized video with the short video identification mark to the key video library:
[0034] Step three: the review extraction unit is used to obtain the composite video within a specified period, and extract the corresponding short video clips from the composite video through identification tags.
[0035] Preferably, in step 2, if the short video only contains a user image, the short video is marked with a custom name corresponding to the corresponding user image;
[0036] If the short video contains a user image and a non-user image, the short video is marked with the custom name corresponding to the corresponding user image plus "unknown name", wherein the non-user image represents an unknown person image that has not been uploaded to the analysis model through the information self-recording unit;
[0037] If the short video only contains non-user images, the short video will be marked with "unknown name".
[0038] The present invention provides a video management system based on a video monitoring platform. Compared with the prior art, it has the following beneficial effects:
[0039] The present invention synthesizes surveillance videos through a business platform, and then obtains short videos containing human images from the surveillance videos, thereby streamlining the surveillance videos, providing convenience for users to quickly view surveillance video data, and effectively improving the efficiency of users viewing surveillance videos;
[0040] The business platform of the present invention further performs custom identification marking processing on all short videos in the composite video, and at the same time sets a playback extraction unit to extract the corresponding short video according to the corresponding custom identification, thereby effectively realizing targeted query of the surveillance video and facilitating users to quickly extract the required video clips. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] Figure 1 is a system block diagram of the present invention;
[0042] Figure 2 It is a flow chart of the implementation of video synthesis of the present invention;
[0043] Figure 3 It is a flow chart of the implementation of video playback of the present invention. DETAILED DESCRIPTION
[0044] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0045] As embodiment 1 of the present invention
[0046] The present invention provides a technical solution: a video management system based on a video monitoring platform, comprising:
[0047] The monitoring platform is used to collect monitoring data captured by infrared cameras in the monitoring area, and then transmit the monitoring data to the business platform, wherein the monitoring data is divided into a number of time period data according to a preset time period, and the time period data is transmitted to the business platform one by one in chronological order, and the time period data includes the monitoring video and infrared information of the same time period;
[0048] The infrared monitoring information is a human signal group obtained at all time points through human infrared sensing technology, and the human signal group includes human signals obtained at each time point, and the human signal includes two signal characteristics of a human signal and a human signal. In this embodiment, the human infrared sensing technology is an existing technology, so it is not described in detail;
[0049] The service platform is used to synthesize the surveillance video and then transmit the surveillance video before and after the synthesis to the media server, and it includes a video processing unit, an information analysis unit, and a video synthesis unit:
[0050] The synthesis process is as follows:
[0051] In the first step, the video processing unit obtains a single time period video in real time, and divides the time period video into several standard time nodes according to the time trend, and the time difference between each adjacent standard time node is the same, and then obtains the corresponding video frame from the time period video according to the standard time node, and obtains the corresponding human body signal in the infrared monitoring information according to the standard time node;
[0052] Then the video processing unit sends all video frames and all human body signals obtained in the video during this period to the information analysis unit;
[0053] Step 2: The information analysis unit imports the video frames and human body signals at the same standard time node into the pre-trained analysis model one by one in chronological order;
[0054] The analysis model obtains the signal characteristics of the human body signal at the standard time point, and determines the human image contained in the video frame based on the face recognition technology;
[0055] At this standard time point:
[0056] If the signal feature acquired by the analysis model is a human signal and the video frame contains a human image, the video frame at the standard time node is transmitted to the video synthesis unit;
[0057] If the signal feature acquired by the analysis model is a human signal and the video frame does not contain a human image, the video frame at the standard time node is transmitted to the video synthesis unit;
[0058] If the signal feature obtained by the analysis model is a non-human signal and the video frame contains a human image, the video frame will not be transmitted to the video synthesis unit;
[0059] If the signal feature acquired by the analysis model is a non-human signal and the video frame does not contain a human image, the video frame will not be transmitted to the video synthesis unit;
[0060] Step 3: The video synthesis unit sorts the video frames transmitted by the information analysis unit according to the chronological order and generates a synthesis sequence library;
[0061] Then, the time difference between two adjacent standard time nodes is obtained and recorded as the comparison difference;
[0062] Then, the continuity of two adjacent video frames in the synthetic sequence library is judged based on the contrast difference. The judgment method is as follows:
[0063] If the time difference between the standard time nodes corresponding to two adjacent video frames is greater than the contrast difference, then there is no continuity between the two adjacent video frames;
[0064] If the time difference between the standard time nodes corresponding to two adjacent video frames is equal to the contrast difference, then the two adjacent video frames are continuous;
[0065] In this embodiment, there is no time difference between standard time nodes corresponding to two adjacent video frames that is smaller than the contrast difference;
[0066] Then, the continuous video frames are obtained from the synthesis sequence table, and the standard time nodes of the first and last video frames are obtained therefrom, and then the two standard time nodes are used as the clipping time periods;
[0067] Then, the corresponding short video is intercepted according to the position corresponding to the interception time period and the video of the time period;
[0068] Afterwards, within a specified period, all short videos of all time periods are synthesized in chronological order to obtain a synthesized video, and then the synthesized video and the original surveillance video are sent to the media server. In this embodiment, the synthesis method of the synthesized video is a prior art, so it is not described in detail;
[0069] The media server includes an original video library and a key video library, wherein the original video library is used to store the surveillance video before synthesis processing, and the key video library is used to store the synthesized video obtained after synthesis processing;
[0070] This embodiment synthesizes the surveillance video through the service platform, and then obtains the short video containing the image of the person from the surveillance video, thereby streamlining the surveillance video, providing convenience for users to quickly view the surveillance video data, and effectively improving the efficiency of users viewing the surveillance video;
[0071] As the second embodiment of the present invention
[0072] See also Figure 1 In this embodiment, based on the first embodiment, the service platform in this embodiment is further used to perform custom identification mark processing on all short videos in the composite video to obtain a composite video with a short video identification mark, and then send the composite video with the short video identification mark to the media server; and it also includes an information self-recording unit, an identification mark unit, and a playback extraction unit;
[0073] Custom identification tags are handled as follows:
[0074] Take a set of short videos as an example
[0075] Step 1: The user uploads the user image through the information self-recording unit, and customizes the user image, and transmits the customized named user image to the pre-trained analysis model;
[0076] The user image is a facial image of the user. In this embodiment, multiple user images can be uploaded. Taking the home monitoring area as an example, the user image can be the facial image of all family members. The custom name is named according to the name corresponding to the facial image of all family members.
[0077] Step 2: The identification and marking unit performs face recognition on all face images in the short video through the pre-trained analysis model, obtains the user image corresponding to the face image in the short video based on the user image after custom naming, and marks the short video, and then sends the synthesized video with the short video identification mark to the key video library:
[0078] If the short video only contains the user image, the short video is marked with the custom name corresponding to the corresponding user image;
[0079] For example, if the short video contains two user images, the short video is marked with “〈***〉+〈***〉”, and “***” indicates the name corresponding to the user image;
[0080] If the short video contains a user image and a non-user image, the short video is marked with the custom name corresponding to the corresponding user image plus "unknown name", wherein the non-user image represents an unknown person image that has not been uploaded to the analysis model through the information self-recording unit;
[0081] For example, if the short video contains 2 user images and 1 non-user image, the short video will be marked with “<***>+<***>+<unknown name>”;
[0082] If the short video only contains non-user images, the short video will be marked with "unknown name";
[0083] For example, if the short video contains two non-user images, the short video is marked with "〈Unknown name 1〉+〈Unknown name 2〉";
[0084] Step 3: The playback extraction unit is used to obtain the composite video within the specified period, and extract the corresponding short video clip from the composite video through the identification mark;
[0085] The business platform in this embodiment further performs custom identification marking on all short videos in the composite video, and at the same time sets a playback extraction unit to extract corresponding short videos according to corresponding custom identifications, thereby effectively realizing targeted queries on surveillance videos and facilitating users to quickly extract required video clips.
[0086] Working principle of the present invention:
[0087] See also Figure 2 , the service platform adds a synthesis timer to regularly check the unsynthesized video data on the system, and concurrently processes the data synthesis through the thread pool. In this embodiment, the regular check is set to an interval of 5 minutes, and the number of threads is configured to 4;
[0088] When submitting a task in the background, a task ID is calculated through the video device national standard number (serial), channel number (channel), channel national standard number (code), start time (start_time), and end time (end_time) to identify a unique task, and a Future is created (used for background checking whether the operation is completed) and put into the cache. Adding Future here facilitates operations when the video is not completed while the user is watching the video.
[0089] See also Figure 3 , add the key video clip table cctv_video_snippet, which is used to save the video clips that users need to review on the business platform:
[0090] CREATE TABLE`cctv_v i deo_sn i ppet`(
[0091] `id` bigi nt NOT NULL COMMENT 'primary key',
[0092] `name`varchar(255)NOT NULL COMMENT'Video clip name',
[0093] `serial`varchar(255)NOT NULL COMMENT'Camera national standard number',
[0094] `channel`varchar(255)NOT NULL COMMENT'Camera channel number',
[0095] `code`varchar(255)NOT NULL COMMENT'National standard number of camera channel',
[0096] `start_time` bigi nt NOT NULL COMMENT 'Video start time in milliseconds',
[0097] `end_time` bigi nt NOT NULL COMMENT 'Video end time in milliseconds'
[0098] `sync_state` int DEFAULT NULL COMMENT 'Synchronization status: 1: Not synchronized to the business system 2: Synchronizing 3: Synchronization completed',
[0099] `video_path`varchar(255)DEFAULT NULL COMMENT'Video storage system path',
[0100] `ct i me`bigi nt DEFAULT NULL COMMENT'Record creation time',
[0101] PR I MARY KEY(`id`)
[0102] ENG I NE = InnoDB DEFAULT CHARSET = utf8mb4COLLATE = utf8mb4_genera l_ci COMMENT = 'Video clip data';
[0103] Among them, sync_state is used to identify whether the business system is synchronized. When the user records key videos, the user sets sync_state to 1.
[0104] The video_path is used to identify the address of the synthesized video on the business platform. This field is empty by default, and the corresponding address will be filled in this field after the synthesis is completed.
[0105] Add a media server, upload the synthesized video files regularly to the media server and return the corresponding address; the user only needs to call up the corresponding video when reviewing the data next time.
[0106] If video synthesis is not performed when the user is viewing, the user only needs to wait for synthesis at that time, without having to wait every time.
[0107] Add a file cleanup timer based on business needs to regularly detect video file data that has been created for more than a certain period of time, such as ct ime, and clean up the files to free up storage space on the media server.
[0108] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.
Claims
1. Video management system based on video surveillance platform, characterized in that: include: The monitoring platform is used to collect monitoring data captured by infrared cameras in the monitoring area, and then transmit the monitoring data to the business platform, wherein the monitoring data is divided into a plurality of time period data according to a preset time period, and the time period data is transmitted to the business platform one by one in chronological order, and the time period data includes monitoring videos and infrared information of the same time period; the infrared monitoring information is a human signal group obtained at all time points through human infrared sensing technology, and the human signal group includes human signals obtained at each time point; A business platform, used for synthesizing surveillance videos, and then transmitting the surveillance videos before and after the synthesis processing to a media server; and it includes a video processing unit, an information analysis unit, and a video synthesis unit. The synthesis processing method is that the video processing unit obtains all video frames and corresponding human body signals of a single time period video, and sends them to the information analysis unit, and then the information analysis unit imports the video frames and human body signals into a pre-trained analysis model for analysis, and then the video synthesis unit obtains the cut-off time period according to the analysis result of the analysis model, and obtains the corresponding short video according to the cut-off time period, and then synthesizes the short videos to obtain a synthesized video, and then sends the synthesized video and the surveillance video to the media server; The specific method of synthesis processing is as follows: In the first step, the video processing unit obtains a single time period video in real time, and divides the time period video into a number of standard time nodes according to the time trend, and the time difference between each adjacent standard time node is the same, and then obtains the corresponding video frame from the time period video according to the standard time node, and obtains the corresponding human body signal in the infrared monitoring information according to the standard time node; Then the video processing unit sends all video frames and all human body signals obtained at each adjacent standard time node corresponding to the video in this period to the information analysis unit; Step 2: The information analysis unit imports the video frames and human body signals at the same standard time node into the pre-trained analysis model one by one in chronological order; The analysis model obtains the signal characteristics of the human body signal at the standard time point, and determines the human image contained in the video frame based on the face recognition technology, and then transmits the video frame at the corresponding standard time node to the video synthesis unit based on the determination result; Step 3: the video synthesis unit sorts the video frames transmitted by the information analysis unit according to the chronological order and generates a synthesis sequence library; Then, the time difference between two adjacent standard time nodes is obtained and recorded as the comparison difference; Then, the continuity of two adjacent video frames in the synthetic sequence library is judged according to the contrast difference, and then the video frames with continuity are obtained in the synthetic sequence table according to the continuity judgment result, and the standard time nodes of the first and last video frames are obtained therefrom, and then the two standard time nodes are used as the clipping period; Then, the corresponding short video is intercepted according to the position corresponding to the interception time period and the video of the time period; Afterwards, within the specified period, all short videos of all time periods are synthesized in chronological order to obtain a synthesized video, and then the synthesized video and the original surveillance video are sent to the media server. The media server includes an original video library and a key video library, and the original video library is used to store the surveillance video before the synthesis processing, and the key video library is used to store the synthesized video after the synthesis processing.
2. The video management system based on the video monitoring platform according to claim 1 is characterized in that: And the human body signal includes two signal characteristics: no-human signal and human-human signal.
3. The video management system based on the video monitoring platform according to claim 2 is characterized in that: The analysis model determines the image of a person in a video frame as follows: Select a standard time node; If the signal feature acquired by the analysis model is a human signal and the video frame contains a human image, the video frame at the standard time node is transmitted to the video synthesis unit; If the signal feature acquired by the analysis model is a human signal and the video frame does not contain a human image, the video frame at the standard time node is transmitted to the video synthesis unit; If the signal feature obtained by the analysis model is a non-human signal and the video frame contains a human image, the video frame will not be transmitted to the video synthesis unit; If the signal feature acquired by the analysis model is a non-human signal and the video frame does not contain a human image, the video frame will not be transmitted to the video synthesis unit.
4. The video management system based on the video monitoring platform according to claim 1 is characterized in that: The continuity of two adjacent video frames is determined as follows: If the time difference between the standard time nodes corresponding to two adjacent video frames is greater than the contrast difference, then there is no continuity between the two adjacent video frames; If the time difference between the standard time nodes corresponding to two adjacent video frames is equal to the contrast difference, then the two adjacent video frames are continuous.
5. The video management system based on the video monitoring platform according to claim 1 is characterized in that: The business platform is also used to perform custom identification marking processing on all short videos in the composite video to obtain a composite video with short video identification markings, and then send the composite video with short video identification markings to a media server.
6. The video management system based on the video monitoring platform according to claim 5 is characterized in that: The business platform also includes an information self-recording unit, an identification marking unit, and a review extraction unit; the custom identification marking processing method is as follows: Step 1: Select a group of short videos, the user uploads the user image through the information self-recording unit, and customizes the user image, and transmits the customized named user image to the pre-trained analysis model; wherein the user image is the user's face image; Step 2: The identification and marking unit performs face recognition on all face images in the short video through a pre-trained analysis model, obtains the user image corresponding to the face image in the short video based on the user image after custom naming, and identifies the short video, and then sends the synthesized video with the short video identification mark to the key video library: Step three: the review extraction unit is used to obtain the composite video within a specified period, and extract the corresponding short video clips from the composite video through identification tags.
7. The video management system based on the video monitoring platform according to claim 6 is characterized in that: In step 2, if the short video only contains the user image, the short video is marked with a custom name corresponding to the corresponding user image; If the short video contains user images and non-user images, the short video is marked with the custom name corresponding to the corresponding user image plus "unknown name", wherein the non-user image represents an unknown person image that has not been uploaded to the analysis model through the information self-recording unit; If the short video only contains non-user images, the short video will be labeled with "unknown name".
