Fast H.264/AVC to HEVC transcoder based on data mining and decision trees

Abstract

High Efficiency Video Coding (HEVC) is gradually replacing its predecessor, the H.264/AVC standard, as the state-of-the-art technology for video compression. However, H.264/AVC has dominated the market for over a decade, so that there is an enormous amount of legacy content that must be migrated. This paper proposes a fast transcoder based on an extensive data mining process on H.264/AVC decoding attributes. The data mining allowed identifying relevant information from the H.264/AVC decoding process, which was conveyed to the C4.5 machine learning algorithm to build a set of decision trees that simplify the complex Coding Unit (CU) size decision in HEVC. Experimental results have shown an average reduction of 44% in the transcoding time, with a small bit rate increase of 1.67%. These results outperform any previous works available in the literature.