At present, China is in the stage of urban population gathering and rapid economic development. The public security situation is grim, and the urgency of security, safety monitoring and security protection is increasing. Public safety technology prevention requires strong technical support and protection. In the handling of emergencies, the detection of criminal cases, and the struggle against terrorism, video surveillance has gradually become an important resource for discovering criminal clues, collecting criminal data, and extracting evidence of crimes. It has become an advantageous means for pre-discipline prevention, incident handling, and post-mortem investigation. . Especially in the process of investigating and cracking the case, video surveillance technology has become an important technical means after public security organs following criminal investigation, technical investigation and network investigation. How to quickly access massive video, how to quickly find valuable clues in the surveillance video, how to automatically identify people and behaviors that endanger public safety, how to quickly locate behavioral trajectories in video cues, and thus enhance the rapid response and coordinated operations of public security organs. The ability to accurately strike and improve the level of social security dynamic prevention and control has become an important task and urgent task for maintaining public safety.
Wang Yulin is a professor and doctoral supervisor at the Department of Information and Electronic Engineering at China University of Mining and Technology (Beijing). Has long been engaged in the research and teaching of smart city construction, automation monitoring, information technology and computer applications. Participated in the formulation of more than 50 industry and national standards, completed 18 scientific research projects at home and abroad, and won two provincial and ministerial level scientific and technological progress awards. He has won two national invention patents, published 5 monographs, and published more than 100 papers, including 22 papers published abroad. At present, he is also a special expert of the Safety Technology Prevention Standardization Committee of the Ministry of Public Security (TC100), an expert member of the Expert Working Committee of the China Automation Society, and a member of the Smart City Expert Committee of the Ministry of Housing and Urban-Rural Development.
It is necessary to carry out research on video image intelligent analysis technology.
In the construction and application of video surveillance systems, there are generally problems of “invisible, incomprehensible, difficult to extract, and slow to retrieveâ€. The main manifestations are that the video images are not clear, the mass information storage is difficult, and the image download speed is slow. Different formats of video can not be played, feature target retrieval is difficult, historical image search comparison error is large, lack of video image information deep mining effective tools and other technical problems. To solve these problems, it is necessary to vigorously carry out research on video image intelligent analysis technology. In the development process of “digital, high-definition, intelligent and networked†in the security industry, Hikvision plays an important role in the development of video image intelligent analysis technology. Especially in the current era of analog surveillance systems, the era of IP era is coming, the value of research on video image intelligent analysis technology is more prominent.
Search and abstract technology is an important part of intelligent analysis of video images
Video retrieval is to retrieve a video containing a certain target according to the target feature, or to find a segment containing the target from the video structure, thereby greatly improving the efficiency of searching and utilizing the video resource. The basis of video retrieval is the target index and content description library obtained after video parsing and feature extraction. Effective video retrieval is achieved by establishing efficient feature descriptions and targets, content indexing libraries, and developing fast and accurate retrieval algorithms. The video summary mainly uses video analysis module, video image feature extraction, feature marking and extraction of video frame elements, reconstruction processing, etc., to summarize the massive video, generate a short summary file, and fuse multiple target elements, with some obvious features. The elements can be quickly retrieved to pinpoint the original video of the target element and find the associated clue video. The search and abstract technology is applicable to surveillance images of various conditions. The retrieval accuracy of HD bayonet images can reach over 95%, and it is also suitable for the retrieval of common standard definition video. Abstract technology is suitable for the analysis and fusion processing of standard definition and high definition video. Currently, search and abstract technologies are basically mature and available. At present, the vehicle feature retrieval of the HD bayonet image is also basically mature, including the retrieval and comparison of the vehicle contour, the license plate, and other parts of the vehicle. The retrieval accuracy of the high-definition image is higher than that of the SD image. Hikvision's Smart2.0 total solution is a powerful reflection of the latest search and abstract technologies.
Video image intelligent analysis has great application prospects in vehicle monitoring
Specifically, in the field of vehicle information extraction and detection, although the related detection and tracking algorithms have been studied for a long time, the application is mainly limited to the acquisition of traffic parameters, and a dedicated detection camera is also required. The identification of the vehicle is also mainly the detection and identification of the license plate. In recent years, the international research on the detection, extraction and identification of more vehicle features has begun, and the technical implementation of quasi-real-time detection of the overall information of the vehicle flow (such as traffic status) has been studied. Obtaining the information of people and vehicles needed from the video, improving the efficiency of video resource utilization, and associating video information with other public security elements has always been a hot issue in video image processing and application, especially in surveillance video. The field of public safety management. Although domestic and foreign research institutions have achieved results on some individual technologies and some products have been put into use, the use of surveillance video is still at the level of artificial leadership. The comprehensive utilization technology of video images needs further development. For example, the video and image resource utilization of the monitoring system such as the bayonet is limited, and the image of the housekeeping license image is not shared with other application systems, and more and more video resources are increasingly brought. Outstanding video transmission, storage issues, and more.
Pedestrian detection and behavior analysis are hot issues in the current research of computer vision and other disciplines.
In recent years, video intelligent analysis products that are available in the market can only detect and analyze common and simple abnormal behaviors such as cross-border, and cannot detect more complicated actions, that is, the applicability to complex scenes is weak. The research on the technique of intercepting the pedestrian's face from the video for portrait recognition is the development direction, but the implementation is difficult, and the hardware and application environment requirements are also very demanding. Therefore, application under normal video surveillance conditions is a new technical challenge.
Due to limitations such as resolution and occlusion, behavioral testing for large-scale populations has always been difficult.
In terms of improving video utilization efficiency, video parsing and retrieval technology has developed rapidly in recent years, but the application system is mainly for the public search engine, not for public security needs and application services. Therefore, based on the current research, combined with public security requirements, further research and development of video application technologies that meet the needs of application services, especially comprehensive video application systems, is an urgent problem to be solved. Due to the rapid increase in the amount of surveillance video, it is necessary to study cloud-based retrieval technology, adopt distributed cloud retrieval data storage scheme, improve the efficiency of retrieval, and retrieve the qualified results in real time for review and query.
In the research and development and engineering application of video image intelligent analysis, Hikvision has always been at the forefront of peers, and hopes that Hikvision will achieve even greater results in future work. It is hoped that the Hikvision new product tour with the theme of “Comprehensively surpassing the simulation and detonating the IP era†can be used as a reference for the development of video image analysis technology in the whole industry.
Photovoltaic Installation Tools
Pv Testers,Photovoltaic Installation Tools,Photovoltaic Tester,Solar Pv Tester
Sowell Electric CO., LTD. , https://www.sowellsolar.com