Research And Implementation Of Video Monitoring System Based On Error Concealment

With the development of the digital home industry, traditional TV programs andservices could not satisfy people any longer, and higher requirement in the digitalhousehold security and intelligent control becomes more and more obvious. Underthis premise and request, network streaming media server, as a representative of theremote network video monitoring system, appears in the video monitoring domain.Video coding standard H.264has been widely applied to video services, owning to itshigh compression ratio, high quality and network adaptability. In the videotransmission, H.264standard can not avoid the symbol error and loss in compressvideo data channel interference, which affects the video quality. Error concealmenttechnology as the video post-processing technology can improve the resistance of thestreaming video error and ensure the quality of video transmission.This thesis is based on the community network project derived from the projectfor the strategic and emerging industries development of Guangdong Province. In thisproject, the smart service provided by digital TV are emphasized. The researchobjective of this thesis is to propose a concealment technology that can be applied tothe video post-processing in video monitoring and control system. We give a briefreview of the video monitoring system as well as the major technologies, andintroduce the key technologies used in the streaming media codec and transmission.This thesis studies the time domain error concealment algorithms, and propose a timedomain error concealment algorithm based on the gradient similarity. This algorithmconceals the error block through the selection of image structure similarity of thehighest alternative macro block, and establishes a more accurate candidatemacroblock matching strategy. Then through the research and analysis of existingstandard concealment order processing procedure, and considering the characteristics of the video monitoring, this thesis propose a moving target detection based adaptiveconcealment algorithm. The wrong macro block is divided into the movingforeground block and the static background block and is concealed with differentmethods respectively, which improves the correctness of video concealment. Theexperiment on H.264test model JM demonstrates the superiority of the proposedalgorithms. Finally, the proposed algorithm is applied to the video monitoring systemon the set-top box decoder module and dealed with the error video frames to improvethe image quality of the video monitoring.