Similarity Measure based 16X16 MB Mode Decision in H.264 Intra prediction

Abstract

A lot of computational power is needed for the intra prediction procedure in the H.264 video encoder. A reconfigurable similarity-based intra prediction method for the H.264 video encoder is thus presented in this paper. The analysis of similarity-based intra-prediction algorithms for H.264 yields a total of five distinct similarity approaches, including sum of absolute differences, sum of squared differences, Hamming distance, Euclidean distance, and cosine similarities. The post-analysis shows that using Hamming distance for similarity-based intra prediction allows for less power and hardware usage, with predicted similarity ranging between 78% and 88%..