HEVC Fast Intra Coding Based CTU Depth Range Prediction

Abstract

Coding tree unit (CTU) partition technique provides excellent compression performance for HEVC at the expense of increased coding complexity. Therefore, a fast intra coding algorithm based CTU depth range prediction is proposed to reduce the complexity of HEVC intra coding herein. First, simple CTU s and complex CTU s are defined in line with their texture complexity, which are limited to different depth ranges. Then, the convolutional neural network architecture for HEVC intra depth range (HIDR-CNN) decision-making is proposed. It is used for CTU classification and depth range restriction. Last, the optimal CTU partition is achieved by recursive rate distortion (RD) cost calculation in the depth range. Experimental results show that the proposed algorithm can achieve average 27.54% encoding time reduction with negligible RD loss compared with HM 16.9. The proposed algorithm devotes to promote popularization of HEVC in realtime environments.