COMPRESSION ARTIFACT REDUCTION IN HEVC USING ADAPTIVE BILATERAL FILTER

Abstract

The High Efficiency Video Coding (HEVC) [8] is the latest video standard developed by Joint Collaborative Team on Video Coding (JCT-VC), a group of video coding experts from ITU-T Video Coding Experts Group and ISO/IEC Moving Picture Experts Group (MPEG). As the demand for HD video (4K and 8K) increased, there is a need for higher coding efficiency than H.264/AVC. Also, there is increased use of parallel processors. So, HEVC [8] has been introduced to support increased video resolution and parallel processing. HEVC obtains about 50% reduction in bit rate when compared to its predecessor H.264/AVC at the same visual quality. All these modern coding standards use block based transforms and coarse quantization to save video bandwidth over a channel by achieving compression. Though HEVC employs in-loop filters such as deblocking and SAO filters to remove compression artifacts obtained due to block based coding and coarse quantization of transform coefficients, there is still a scope for improvement. As in-loop filters are part of the standard, any modification of in-loop filters would modify the coding standard. So, post-processing techniques have gained popularity as they would not disturb the existing standard and reduce compression artifacts. The thesis focusses on applying bilateral filter on HEVC decoded frames adaptively to reduce compression artifacts and still maintain very good visual quality and take less time to apply the filter than the in-loop filters. PSNR (Peak Signal to Noise Ratio) metric is used to evaluate subjective quality of the video.