Design And Implementation Of B/S Mode Video Surveillance Client Based On JavaEE

With the rapid improvement of embedded technology, multimedia technology and Internet technology, the video surveillance industry has achieved great development in recent years. It has been applied to all walks of life and even ordinary family, and will continue to develop to the direction of high-definition and intelligentization.In traditional video surveillance system, the client software generally uses the C/S mode to develop but it has trouble to upgrade and high maintenance cost. With the development of web technology, the B/S mode has been more and more popular. It has become the dominant mode in client development. In this paper, we design and implement a B/S mode video surveillance client based on JavaEE technology in a practical video surveillance system, complete the functions of processing user requests, video session establishment, receiving, decoding and rendering video data. The main contents are as follows:1. Analyzes the main technology needed in B/S mode video surveillance client, including JavaEE technology, YUV color space used for video rendering, H.264 encoding used for video compression, RTP used for video data transmission and SIP used for video session establishment.2. According to the specific requirements of video surveillance client, we devide the entire client software into several functional modules and selects appropriate technical framework to realize each module. In detail, this paper uses MjSip to realize SIP communication module, uses Struts2 to realize user request processing module, uses Jrtplib to realize the receiving of RTP packet with video data, uses ffmpeg to realize the decoding of H.264 data and uses SDL2 to realize the rendering of YUV video frame after decoding.3. In order to solve the image detail fuzzy problem caused by traditional bilinear interpolation algorithm, we propose an improved algorithm which uses threshold method to divide the image into approximate edge region and non-edge region according to the standard deviation of grey value of the 4 neighbor points, for the pixels in non-edge region it directly uses bilinear interpolation and for the approximate edge region it uses the method of making weighted fusion of both nearest neighbor and bilinear interpolation algorithm. This paper descripts the improved algorithm in detail and implements it for video frame of IYUV format, finally evaluates the performance of the algorithm in both subjective and objective perspective.