Dependency channel model for transform domain Wyner-Ziv coding

Distributed video coding is a new video coding paradigm aimed at shifting the computational expensive motion estimation to the decoder. Side information is generated at the decoder as a prediction of the current frame. This creates a virtual dependency channel between the source video at the encoder and the side information at the decoder. Error correcting channel codes are used to correct the side information. Many recent distributed video coding systems are based on turbo codes. While the performance of turbo codes can approach the Shannon bound when used on standard noisy network communication channels, it is not obvious that they can deal with non-stationary noise encountered in distributed video codecs. In this paper, we propose a channel model and show how the accuracy of this model affects the performance. Keywords— distributed video coding, Wyner-Ziv coding, turbo code, virtual dependency channel, noise model