Analysis of CABAC context models efficiency in the HEVC

摘要

This paper presents detailed analysis of the CABAC efficiency when operating in the HEVC in a number of bits per number of bins manner. The paper reveals what fraction of bit is produced by the CABAC for each and every binary symbols encoded in HEVC. The statistics of bins encoded with a given efficiency are presented for the three types of video data: 1) prediction error of image samples, 2) motion data, and 3) control data.