Design and implementation of distributed video coding architecture

摘要

The present day video coding standards such as MPEG and H.264 are mainly used for broadcast applications, where the encoder is more complex than the decoder. However, for applications such as low power surveillance networks and wireless video cameras, where there is a dearth of hardware at the encoder, the encoder needs to be simpler, but the decoder can be complex. Distributed Video Coding (DVC) technique is an emerging coding paradigm for the above kind of applications. The primary objective of this paper is to design and implement Distributed Video Coding technique, with high compression efficiency and resilience to transmission errors. The main concern of the design is to shift the computationally complex part in the encoder, i.e., Motion Estimation (ME), to the decoder, thereby making the encoder less complex. The whole design is tested on MatLab software and Wyner-Ziv (WZ) part of the codec is implemented on Beagle Bone Black hardware. The test results show that the performance of the DVC Codec developed is close to the performance of H.264 conventional codec, which is a splendid result according to Slepian-Wolf and Wyner-Ziv theorems. A PSNR value of more than 35 dB is achieved along with a compression ratio of around 7:1. Also, there is more than 50% reduction in the encoding time when compared with standard codecs like MPEG, H.264, which was the main concern of this paper.