Fusion of global and local side information using Support Vector Machine in transform-domain DVC

Abstract

Side information has a strong impact on the performance of Distributed Video Coding. Commonly, side information is generated using motion compensated temporal interpolation. In this paper, we propose a new method for the fusion of global and local side information using Support Vector Machine. The global side information is generated at the decoder using global motion parameters estimated at the encoder using the Scale-Invariant Feature Transform. Experimental results show that the proposed approach can achieve a PSNR improvement of up to 1.7 dB for a GOP size of 2 and up to 3.78 dB for larger GOP sizes, with respect to the reference DISCOVER codec.