An H.264 weighted prediction parameter estimation method for fade effects in video scenes

This paper presents a parameter estimation method that makes H.264´s weighted prediction (WP) practical for fade effects with quasi-optimal accuracy and low computational cost. Its high accuracy results from the fact that it is based on a theoretically-derived approximate relationship between ideal WP parameters and image characteristics that are robust with respect to motion. The method uses only simple intra-frame image characteristics, and its computational cost is therefore negligible compared with that of the encoding process itself. Experimental results show that its performance is only 1.6% worse than when ideal WP parameters are used, as opposed to degradation of from 56% to 73% with conventional methods.