Power-aware scheme for motion estimation

Abstract

Video compression has gain major importance with the use of video call, video transmission and streaming on wireless mobile devices. Video compression algorithms had been implemented mainly in software. However, the battery life restrictions have encouraged the hardware implementation of these algorithms. The present work is a power-aware scheme that take advantage of the temporal redundancy in the video stream as well as the status of the battery. The introduction of a simple motion detection algorithm decreases the need of calling the motion estimation and compensation process. The calls of the motion estimation and compensation process reduces operations causing a dynamic power reduction and a speed-up of the coding process. While a power-aware scheme preserve battery life by preventing a subdivision of the macroblock considering motion detection and battery capacity. The reduction on execution time for MPEG4 is in the range of 9% to 22% while achieves an average power reduction of 88%. While H.264/AVC achieves an execution time in the range of 99% to 89% with an average power reduction is 98%.