Chinese Journal of Computers   Full Text
Title	Error Resilient Video Transcoding Based on Layered Rate Distortion Optimization
Authors	HAN Zhen1) HU Rui-Min 1),2) CHEN Hao2) LI Ming2)
Address	1)(School of Computer, Wuhan University, Wuhan 430072) 2)(National Multimedia Software Engineering Research Center, Wuhan University, Wuhan 430072)
Year	2009
Issue	No.6(1152—1161)
Abstract &   Background	Abstract Error resilient video transcoding can insert error resilient tools in the compressed video to enhance error resilience of the video over wireless channels. However the error resilience insertion increases bit rate while reducing distortion. So the rate distortion optimization is required. In order to solve this problem, this paper proposes an error resilient video transcoding algorithm based on layered rate distortion optimization. This algorithm improves the flexibility by performing rate distortion optimization with different granularity for synchronization marker insertion and intra macroblock refresh on the frame layer and macroblock layer respectively. Moreover this algorithm improves the accuracy by accounting for the intra refresh, intra prediction, motion vector prediction on the frame layer and reduces the complexity by reducing the candidate macroblock modes on the macroblock layer. The simulation results show that the proposed algorithm improves PSNR by 0.6~1.1dB and reduces the complexity by about 25% compared with the algorithms based on single layer rate distortion optimization. Keywords video transcoding; error resilient; rate distortion optimization; synchronization marker insertion; intra macroblock refresh Background The work in this paper is one of the series studies of Adaptive Wireless Video Transmissions in the National Multimedia Software Engineering Research Center. The main objective of the project is to establish effective schemes to translate video adaptively through wireless heterogeneous networks. As a key technology to solve this problem, wireless video transcoding can make the video content adaptive to the capabilities of diverse networks and client devices. Thus, it is one of the essential components for current and future multimedia systems that aim to provide universal access in a wireless heterogeneous network environment, such as mobile e-business, mobile entertainment, mobile surveillance. The center has worked in the wireless video transcoding including rate control and mode decision for transcoding since 2005. These works are supported by National Natural Science Foundation of China (No.60472040) and Innovation Plan of Ministry of Public Security (No.2005YYCXHBST114). As one of key technologies of wireless video transcoding, error resilient video transcoding can insert error resilient tools in the compressed video to enhance error resilience for video transmission in a wireless heterogeneous network environment. However the error resilience insertion increases bit rate while reducing distortion. So the rate distortion optimization is required. In order to solve this problem, researchers have proposed some error resilient video transcoding algorithms based on single layer rate distortion optimization. However the flexibility, accuracy and complexity of these algorithms still need to be improved. So this paper proposes an error resilient video transcoding algorithm based on layered rate distortion optimization. This algorithm improves the flexibility by performing rate distortion optimization with different granularity for synchronization marker insertion and intra macroblock refresh on the frame layer and macroblock layer respectively. Moreover this algorithm improves the accuracy by accounting for the intra refresh, intra prediction, motion vector prediction on the frame layer and reduces the complexity by reducing the candidate macroblock modes on the macroblock layer. The simulation results show that the proposed algorithm improves PSNR remarkably and achieves the best PSNR compared the reference algorithm while maintaining low complexity. The next step is to introduce the novel error resilient tools adopted by H.264 in the video transcoding scheme.