I. Introduction
At present, the video reconnaissance mode is mainly based on manual review, supplemented by simple intelligent video analysis. Professional watchers are still required to conduct 24-hour video patrols or to browse and query a large number of videos based on the event information. The work mode based on manual intervention has gradually become the bottleneck of video applications. With the explosive growth of video resources, the low application performance of massive video resources has begun to restrict the development of video service applications. At this time, the construction of security big data applications has been put on the agenda. This year, the security industry leader Hikvision The concept of SDT security big data is also proposed.
In the two sessions of 2015, Premier Li Keqiang clearly stated the big data construction action plan in the government work report, and big data has become a national basic strategic resource. Big data application is an inevitable requirement for realizing the construction of intelligence-led policing mode, and it is an inevitable development trend of public security information construction. In the construction of security big data applications, using advanced technologies such as video structured description technology and intelligent image analysis technology, structured data extraction for people, vehicles and moving targets in massive video will be massively unstructured video and semi-structured. The transformation of images into valuable structured information and the realization of "video analysis" will be an important basis for public security organs to realize the application of security big data applications. In this paper, the author will talk about the new changes brought by video analysis to video surveillance based on the current market environment and his own experience in the security industry.
Second, what is "video resolution"?
What is "video resolution"? Video parsing is a technique for synthesizing textual information that can be recognized, understood, and retrieved by computers and people by using semantic analysis of video content, using object detection, feature extraction, object recognition, and deep learning. The video parsing method can be divided into three types at present: the first one is the traditional “pure manual†parsing mode. The policeman conducts the human feature annotation on the suspect target while reviewing the video, which is convenient for later retrieval calls. This method is less efficient. It is no longer able to meet the needs of massive video analysis at this stage; the second is “man-made, smart-assistedâ€. Due to the advent of front-end intelligence and back-end intelligent resolution devices, the initial analysis of the original video can be performed to a certain extent. The liberation of manpower, improve the efficiency of analysis; the third is the "smart-based, artificial-assisted" analysis mode, with the development of intelligent analysis equipment, through video structured analysis and big data, cloud computing and other technologies to achieve video images Human resources have also been gradually replaced by structured analysis of resources, rapid retrieval, and big data comparison collisions. However, due to the fact that smart devices do not have strict normative guidance for the specific resolution of suspected targets, and for more complicated cases, the content of the analysis cannot meet the demand for clues in the case, so at this stage, manual analysis is still needed. The analytical method of “mainly and artificially assisted†is widely used at this stage.
This paper combines video detection with the third analysis method widely used at this stage, and explores the system construction model of “offline analysis system + centralized analysis system + manual analysis assistanceâ€. The offline analysis system can be directly passed through the crime scene. The offline forensic equipment analyzes the video and pictures involved, and the formed clue information can be directly judged by the user on the spot, which can improve the efficiency of the case to a certain extent. The centralized analysis system receives the original video and pictures transmitted by the front-end camera, and can perform real-time and historical video analysis of the human, vehicle, and activity target clouds, and perform secondary information recognition on the card-passing picture, and finally pass the standard external service interface. The public security business provides structured basic data, which is mainly applied to the analysis and application of massive video in large cases. Finally, for some major cases, major cases and special cases, manual analysis assistance is needed. When the video is viewed, the investigators directly describe the suspected targets, form clue information, assist the offline analysis system and the centralized analysis system to complete the lead screening. And integration.
Third, the new mode of video investigation
"Video Resolution" provides powerful data support for video surveillance and lays the foundation for the exploration of new models of video detection. Based on the "offline analysis system + centralized analysis system + manual analysis assistant", this paper proposes a video detection mode that is different from the traditional manual intervention and supplemented by intelligence, and according to the type, size and severity of the case. Differentiating and exploring new video detection modes include: individual combat mode, cluster operation mode, mobile detection mode, centralized mode, video detection studio mode and synthetic combat six modes, which can be satisfied by different video detection modes. The video investigation work of different scenes and different needs provides a certain idea for the public security to solve the case.
1. Individual combat mode
When a general case occurs, the investigator can bring the video survey equipment and the offline forensic equipment to the crime scene to understand the case, collect the video of the crime scene and the surrounding area, and conduct target screening through the “offline analysis systemâ€. The formed structured data can be used for on-site judgment by offline forensics equipment, quickly locate suspect targets, and complete video investigation work. The individual combat operation mode can meet the normal business needs of the general case detection needs.
2. Cluster combat mode
When an emergency case occurs, multiple off-line investigators can be assigned to bring multiple offline forensics equipment to the crime scene to understand the case, and multiple offline forensics equipment will be formed into a local area network to form a video investigation task group, multi-person collaboration, and resource sharing. Co-complete the video collection of the crime scene and surrounding areas, and conduct target screening through the “offline analysis system†to form structured data, provide data support for on-site research and judgment, and improve the work efficiency of the police. The cluster combat mode is mainly applicable to cases where the case is more complicated and the time is more valuable.
3. Mobile detection mode
When a case with a complex scene occurs, such as when the investigator is inconvenient to arrive at the scene, or if a covert investigation is required, the investigator may carry the mobile surveillance equipment for temporary control. The video involved is transmitted back to the centralized analysis through the operator VPN private network. The system "stores and parses, and conducts remote research and judgment through the actual application platform to complete the video investigation work. Mobile reconnaissance equipment is an important complement to conventional fixed video surveillance.
4. Case mode
When there are major cases and major cases, the investigation police can rely on the social security monitoring network sharing platform, video survey equipment, and video mobile emergency equipment control equipment to conduct online download and offline upload of the crime scene and surrounding video. The centralized analysis system and the manual analysis assistant method initially screen the video involved, form a case clue, use the case comprehensive combat platform, and perform clue integration on the parsed structured data and the original video, and associate the “human and vehicle†elements. Multi-person collaboration, time-space trajectory research, case serialization, comprehensive collision, etc., comprehensive research and judgment to determine the direction of the next investigation, or to generate a settlement report after the case is closed. The centralized mode relies on the video analysis center to improve the lead conversion efficiency and improve the detection rate. Finally, the centralized mode needs to pay attention to the information management, query, sharing and storage of live video cases. Provide strong data support and guarantee for the comprehensive utilization of the video information involved in the case and case, and the case.
5. Video Detection Studio Mode
The so-called video reconnaissance studio, in accordance with the requirements of "reconnaissance, forensics, image processing, analysis and judgment", recruits experienced investigators and imaging technicians to form professional investigation teams, and based on video detection business processes and application systems. The deployment requirements are to configure the corresponding video surveillance workplace and work space for the investigation team. Through the work space planning, the investigation personnel can work independently or jointly. The overall workflow of the video investigation studio is similar to the centralized mode of the case. From the video capture à video review à video analysis à video research judgment, the video analysis here is similar to the centralized mode of the case, but also through the "centralized analysis system" plus "manual analysis assistance" The main difference is that the analysis of the video reconnaissance studio mode emphasizes “special area special person†and “special person-onlyâ€, establishes a video investigation professional team, and enhances the special skills of video reconnaissance personnel. The video reconnaissance studio layout is as follows:
6. Synthetic combat mode
Establish a modern police service mechanism of "combined warfare" in the "four-in-one" simultaneous investigation of criminal investigation, technical investigation, economic investigation and network investigation. When a criminal case occurs, the police officers quickly gather and organize the information involved, the first time Synchronize the case, and carry out video analysis and comprehensive research around the clues. This kind of synthetic combat mode can maximize the integration of professional technical resources of various police types, and rely on video surveillance network sharing system, social resource management platform, bayonet subsystem, practical application platform, etc. to maximize the accumulation of these professional technical resources to the investigation and resolution. To achieve coordinated collaboration and precision strikes of resources, thereby improving the overall effectiveness of the strike and comprehensively improving the core combat effectiveness of the public security organs.
Fourth, the conclusion
The era of SDT has arrived. The application of security big data to video detection and detection has become an inevitable trend. The analytical system model of “offline analysis system + centralized analysis system + manual analysis assistance†is constructed, and the structured data of “processing and production†is established and established. People, vehicles, and articles are focused on the target video image information clue suspicion database, case events, inventory, and other types of thematic business databases, providing video image information, feature information and other data support for public security business big data applications, thus making video In the context of big data, the investigation fully exerts the role of delineating the direction of investigation, narrowing the scope of investigation, locking suspects, and even identifying crimes.
Delivery from Malaysia Warehouse
Guangzhou Fengjiu New Energy Technology Co.,Ltd , https://www.flashfishbattery.com