Fast Computation Algorithm of Discrete Cosine Transform for Versatile Video Coding

Abstract

Video Compression is one of the most essential tasks in applications such as video newlinebroadcasting, video communications, and video streaming. With the increase in 4k and newlinebeyond resolutions, there is a huge demand for high quality videos with high compres- newlinesion efficiency. There are many video coding standards and Versatile Video Coding newline(VVC) is one of the latest and most advanced video coding standards developed by newlineITU-T and IEC/ISO. VVC is aimed to increase the compression efficiency of videos by newlinetwo times compared to its predecessors (like High Efficiency Video Coding or HEVC) newlineat the cost of the increase in the encoder complexity. There are many blocks/modules newlinein the VVC encoder and transform block is one of the critical blocks in the encoder, newlinewhich consumes almost 40% of the total encoding time. This thesis proposes a novel newlinefast transform algorithm that reduces encoder time complexity without compromising newlineon output video quality and compression efficiency. newlineThere are many transform algorithms for video encoding and Multiple Transform newlineselection (MTS) algorithm is one of the fast transform selection techniques which can newlinebe used in VVC. It employs various types of Discrete Cosine Transform (DCT) and newlineDiscrete Sine Transform (DST) algorithms by selecting them adaptively based on the newlinestatistics of the video content. This thesis proposes a novel fast MTS algorithm and the newlineexperimental results show that there is an overall reduction of 22% of total encoder time complexity with negligible loss in output video quality and compression efficiency newline