Research On Smart Video Surveillance System

Video surveillance is an important part of the security system. With the rapid development of pattern recognition, image processing and computer vision technology, video surveillance system has been used more and more widely. As the enlarging of video surveillance area, and the monitoring time is usually around the clock, thus producing a huge amount of monitoring video. When the video data is used for post-mortem analysis, it will waste a lot of time in searching the relevant information. The traditional monitoring mode that completely relying on manual is boring and time-consuming. Nowadays, we need to move our concentration from the saving of monitoring data to automatically analysis of the data in real time, which is able to send an alert message when an accident happens. Intelligent monitoring system is not have the ability of real-time monitoring of various potential threats, but also helps in filtering out a lot of useless information, and able to searching the needed information quickly and effectively.This paper designs and implements an intelligent video surveillance system, which can receive the video information from multiple cameras in different monitoring points, and display the monitoring scenes in real-time. The remote control center of the system can detect the moving objects in real-time, the hardware platform of our system which is also the web server could communicate with the remote center bidirectional.This study is focus on the moving object detection and the transmission of image data in the intelligent monitoring system. By detecting the real-time moving objects in the monitoring scene, we could able to judge if the current frame is in motion or static, and the judging could help us filtering out static frames in the while of saving the monitoring data. Thus not only greatly reduces the needs of storage space of the surveillance video, but also improves the efficiency of the post-mortem analysis. In general, the monitoring data from a camera will be transmitted to one remote monitoring center, if there are a lot of cameras that installed here and there to covering the monitoring area, the monitoring data that needed to be transmitted to the control center would be very huge, that is easy to cause an network congestion, which will affect the real time display of monitoring scene. To avoid such problem, some mechanisms are used to control the data size that transmitted on the internet at a time. On the server side, standard JPEG quantization table and Huffman table is used in the compression of image data, and the compressed data is transmitted with UDP protocol which is faster than TCP protocol. As we got exactly the same JPEG header for each image, only the data part of a JPEG image will be transmitted, as the remote control side received the image data, the JPEG header will be added to form a whole image. The web server also receives feedback from the remote control center, and be able to adjust the data sending speed according to the feedback information.The paper is organized as follows. The backgrounds and significance of this research is given in the first part, which briefly introduces the development of video surveillance system. The second part analyzes the overall design of the system. In the next part, we introduced the data transmission mechanism, image data encoding and decoding algorithm, and moving object detection algorithms. Next, the realization of the intelligent video surveillance system and the analysis of the system performance are described in detail. Conclusion is addressed at the end of this paper.