Research And Implementation Of Intelligent Network Video Surveillance System Supporting Auto-focus

Along with the improvement of people's living standards and the awareness of security, the video monitoring as a means of combating crimes and guaranteeing the order of people's daily life, should be with a wide range of applications and very broad market prospects. This paper attempts to analyze and research the key techniques of video surveillance, and adopt DSP+ARM to develop an intelligent network video surveillance system.This paper makes image focusing to the motion regions on the basis of motion detection and with Laplacian gradient function as an evaluation function. In order to improve the sensitivity of the image focusing, an adaptive threshold T is introduced to the former evaluation function. The results show that the sensitivity of the focusing evaluation function is improved observably.For convenience of the video recording retrieval for the clients, this paper analyzes the method of video retrieval and hereby brings forward a technique of video label, which takes the motion image as an event label and then find out the position of the event in the video file accordingly. Being simple and applicable, this method can save users'time for browsing the video file and improve researching efficiency.In order to improve the real-time transmission of video streaming, the paper uses RTP/RTCP protocol to transmit the video streaming and makes experimental comparison between the transmission performance of TCP/IP protocol and RTP/RTCP protocol respectively in the condition of network light load and heavy load. The experiment shows that RTP/RTCP protocol is more suitable for the transmission of video streaming.At the end of the paper, a prototype system is achieved by the realization of key techniques. The experimental tests prove that the function of the system has been reached as expected.