Skip to Main Content
We consider here a low-complexity approach to the problem of multi-view distributed video coding. Specifically, we assume that motion-JPEG is used as the base coding algorithm and analyze the effects of factors such as additive noise and region misalignment on the conditional coding problem. Our goal is to characterize the achievable compression gain for a practical coding scenario where block-wise frame correlations must be estimated on-the-fly at the encoder in a highly efficient manner. We show that additive noise has a greater impact on the performance of a distributed video coder than pixel domain shifts.
Date of Conference: 23-25 May 2010