JND-based Wyner-Ziv Video Coding

Distributed video coding (DVC) is the implementation of distributed source coding in video coding. Its core idea abide by the fundamental of distributed source coding, where correlated sources are encoded independently but decoded jointly. Video coding is meant to achieve the best possible reconstruction quality for a given bit rate. In recent years, research community think highly of DVC and make their work focus on how to attain a better DVC system. In this paper, we introduce the concept of Justnoticeable distortion (JND) in distributed video coding system, a measure of maximum image distortion that the human eye cannot detect, which can effectively improve the encoding efficiency and not affect perceptual quality. Meanwhile, the quality of the side information (SI) plays an extremely significant role in the performance of distributed video coding system. The better the quality of side information is, the higher the system performance will be. Also, experimental results in this paper illustrate that by using one side information distributed video coding system with JND model can achieve the same effect as multiple SIs system without JND model.