Research And Design Of Wyner-Ziv To AVS Video Transcoder

AVS1-P2 is the newest video coding standard of Audio Video coding Standard (AVS) workgroup of China, which adopts the motion-compensated hybrid coding framework. In such a framework, intra and inter predictions are performed at the encoder side to remove the spatial and temporal correlations. Given that predictive coding uses motion estimation (ME), which is a computation-dense process, the video encoder is typically more complex than the decoder. This architecture is much suitable for applications where the video sequence is encoded once and decoded many times, such as broadcasting or video-on-demand. However, due to the limited energy and computing ability, several emerging applications such as wireless video surveillance, low-power video sensor networks and mobile camera phones can not afford such a complex encoder.Distributed video coding (DVC) is a new video coding paradigm, which performs intra-frame encoding and inter-frame decoding to shift the complexity from encoder to decoder. It is a promising technology to compress video source obtained by mobile multimedia devices, such as phone camera and wireless video sensor networks.AVS fits well some low-decoding complexity applications, while DVC is suitable for applications requiring low-encoding complexity. However, in some applications, low complexity is simultaneously needed at both encode-decoder sides. Mobile video communication is an example. For such scenario, both AVS and DVC do not work. One solution is to use a transcoder from a DVC stream to an AVS one.In this paper, our goal is to get the best of both worlds. We propose a fast video transcoder from Wyner-Ziv to AVS. The transcoder design aims at reducing encoding complexity of AVS. Taking into account that mode decision is very time-consuming, in the proposed scheme; we discard the mode decision and directly re-use the corresponding decision results in the process of side information generation. At the same time, motion estimation in AVS encoder is speeded up by using the derived motion vectors in Wyner-Ziv decoder as searching origins. Then motion vector refinement is performed to further improve the accuracy of motion estimation in the transcoder.