Energy efficient motion estimation using stochastic processing for video conferencing applications -

Abstract

One of the serious challenges facing smartphones and other portable devices nowadays is battery life. We aim at improving the energy efficiency of several multimedia applications in order to improve the battery life of a mobile device while still maintaining an acceptable level of user experience. In this work, we first design a stochastic processing technique for the JPEG encoder in order to limit its energy consumption particularly on mobile devices. We then present the full profiling of the state-of-the-art video encoder HEVC. A similar procedure is then followed where stochastic processing is applied on the motion estimation stage of the HEVC encoder in order to reduce its energy consumption. Several tests were performed on standard video sequences. The quality of the output videos was tested both subjectively and objectively using the standardized video quality metric VQM. The results for both the JPEG and HEVC encoder are presented in the work below. Images and video snapshots are shown along with the relative percentages in energy savings. We demonstrate the feasibility of using stochastic processing for multimedia encoders in general and analyze the energy savings and their effect on the end user.