A genetic frame fusion algorithm for side information enhancement in Wyner-Ziv video coding

This paper presents a novel frame fusion algorithm that aims at improving the side information in distributed video coding. Using a genetic algorithm, the proposed technique combines several interpolated frames, obtained by any existing methods, in a fusion based approach. Simulation results show a significant improvement in the side information quality, where the gain in PSNR exceeds 1dB most of the time, and can even reach 20 dB for some frames, compared to classical interpolation techniques available in the literature.