Multiple Description Scalar Quantization-based Distributed Video Coding

With the development, innovation and continuous evolution of wireless multimediatechnology, multiple description distributed video coding design concept completelysubverts traditional technical solutions such as distributed video coding and becomes anadvanced theory. Because of high efficiency and robust coding technique, multipledescription distributed video coding is suited for resolving unreliable-transmit problemcausing by network congestion and abnormity.This thesis introduced the distributed video coding (DVC) and designed a multipledescription DVC for the packets loss problem of unreliable wireless transmission. InDVC, the side information quality which is estimated by the decoder has a great impacton the performance of system. The side information is obtained through the estimationof adjacent key (K) frame. Reducing the packet loss rate of transmitting K frame canimprove the quality of K frames and decoded images. This thesis analyzed the ratedistortion performance of multiple description and the impact of complementation andenhancement among each description on decoded K frame. This thesis integrated theclassical multiple description scalar quantization (MDSQ) method in the key frame oftraditional distributed video coding. In addition, a new standard of image evaluationwhich is called structural similarity (SSIM) was introduced in the design of decoderverdict mechanism. Programming achieved the overall multiple description scalarquantization coding based DVC (MDC-DVC) system. Compared with the traditionaldistributed video coding system, simulation results showed that the performance ofMDC-DVC system has been significantly improved under different packet loss rate.