Abstract
Distributed Video Coding (DVC) is a new coding paradigm suitable for extremely low complexity video encoders. This makes it suitable for limited resources applications such as sensor network. Wyner-Ziv video coding, the most representative method in DVC uses channel coding, such as turbo code or LDPC code, to reconstruct a frame based on side information. It is important to transmit just right amount of channel coded data, and this problem has been generally tackled by using a feedback channel.
In this paper, we propose a hybrid distributed video coding approach which combines motion compensated prediction and Wyner-Ziv video coding. This allows to accurately estimate the required channel code rate, thus eliminating the feedback channel. Experimental results indicate that the proposed method achieves good performance while maintaining low complexity.