Research And Implementation Of Video Surveillance System And Its Related Technologies

With the rapid advance in network and multimedia technology, as well as the increasing need of social security, video surveillance technology has been widely used for police, airports, banks and many other fields. This demonstrates that video surveillance is a technology with broad space for development and market prospects.This research topic comes from a project financed by China Mobile Corporation. The project aims to design and implement a third generation video surveillance system, with advanced features such as remote monitoring centralized management and mobile monitoring. The further goal of the project is to create a city-level video surveillance platform providing high quality public video surveillance service. Most of the conventional video surveillance systems adopt simple overlay network topologies that are subjected to high workload problem in the sever side, resulting that the system can not support large concurrent user. In addition, various illuminate condition, hardware, network and other factors lead low video quality, especially in night. It's an important issue to support large number of concurrent users under limited bandwidth. While improving the video quality especially the night one is of great importance to video surveillance systems.This work design and implement a video surveillance system named E-Vision, attempting to provide surveillance service to large number of concurrent users and improve the video quality at night. We make use of J2EE development technology such as Struts, Spring and Hibernate to implement a centralized management platform based on the overall design of the video surveillance system. This platform integrates management functionalities such as user management, device management, video sources management, alarming management and logging management. To meet the need of providing surveillance of large number of concurrent users, the work make use of the correlation of users interested in different video channels to propose a novel application layer multicast model based on the model of DONet. To address the low video quality problem at night, we propose an effective image enhancement algorithm that combining the advantage of histogram equalization and wavelet transformation.The proposed'E-Vision'surveillance system has been successfully deployed on China Mobile Company and The Kindergarten of Zhongshan. These practical applications show the efficiency of this work.