Distributed video coding for resource constrained video applications

Video coding technology has played a key role in the explosion of current multimedia society with increasing resolution and quality. Such big success is largely built on the conventional video coding paradigm where motion estimation and compensation are performed at the encoder. This asymmetry in complexity is well-suited for the applications where the video sequence is encoded once and decoded many times. However, some new emerging applications such as wireless video surveillance, wireless PC cameras and multimedia sensor networks require a low complexity encoding, while possibly affording a high complexity decoding. Therefore, a challenging problem emerges with the new type of visual communication system is how to achieve low complexity encoding video compression while maintaining good coding efficiency.;Distributed video coding (DVC) provides low complexity encoding solutions for video communication with limited computational power or energy constraints. In DVC, the source video information is independently encoded at lightweight encoders. At the decoder, all the received bitstreams are jointly exploited their statistical dependencies between them. In such a way, motion estimation and its computational complexity is shifted from the encoder to the decoder. However, DVC also has its own restrictions. The low coding efficiency remains a challenging issue for DVC compare to the conventional video coding. Although DVC is robust to channel loss due to its intrinsic feature of independent encoders and joint decoder, the error resiliency for medium to large transmission errors is weak.;In this dissertation, previously proposed low-complexity DVC (LC-DVC) architecture is firstly introduced. After that, a continued work is presented to further improve quality of SI. The proposed method is called spatio-temporal joint bilateral upsampling (ST.JBU) based Si generation, where geometric closeness of pixels and their photometric similarity is exploited to reduce the noise while preserving the edge information. Moreover, a distributed multiple description coding (DMDC) scheme is proposed by combining the multiple description (MD) coding into LC-DVC to improve its error resiliency. All the proposed schemes are well described and the rate-distortion analyses are presented in this dissertation. All these features have made the LC-DVC a great solution for resource constraints applications.