A novel architecture for parallel multi-view HEVC decoder on mobile device

Abstract

The multi-view HEVC (MV-HEVC) extension was finalized in July of 2014 by the Moving Picture Experts Group and the Video Coding Experts Group. Recently, multi-view videos based on stereo representations are becoming widely popular. Also, a variety of multimedia contents are now available for mobile devices. A real-time multi-view video decoder is therefore needed. In mobile devices, a real-time decoding multi-view video is difficult because of the increasing number of views, spatial resolutions, and limited speed of the processors on mobile platforms. In this paper, we propose a novel architecture for a real-time decoder in mobile devices. The proposed MV-HEVC decoder uses parallel-optimized multi-view video decoding with multi-threading, using advanced reduced instruction set computer machine (ARM) Cortex multi-core processors. Moreover, it is optimized in single instruction multiple data for an ARM platform. The proposed multi-core decoding architectures enable multi-threading with minimum processing overhead. Experimental results show that the proposed multi-view video coding increased the speed by around 2.4–4.8 times in the ARM platform compared to MV-HEVC.