Performance Evaluation of Next Generation Video Compression Approaches Using Objective Quality Measurement Methods

Abstract

With the expansion of services based on television and video transmission over the Internet, the need for more effective compression of video has become important. Since all of the video compression approaches in use are based on lossy compression, some degree of quality loss is inevitable in the compression process. Instead of subjective methods for managing this quality loss, video quality should be measured automatically and objectively. The recently developed VP9 and AV1 approaches, which are currently in the expansion and experimental phase, offer better compression levels than the currently widely used H.264 (AVC) approach. On the other hand, the hardware requirements of these new approaches are higher than those of AVC. In this study, results of the laboratory tests comparing AVC, VP9 and AV1 approaches for different video types in terms of 'compression ratio', 'quality level' and 'compression time' where quality loss is automatically determined based on VMAF, PSNR and SSIM values.