HEVC-based adaptive quantization for screen content videos

In general, quality of compressed videos for screen contents is not as good as those for natural videos with the conventional video compression standards due to different properties of textures. In this paper, we present an adaptive quantization method on high efficiency video codec (HEVC) in order to improve coding gain for screen contents. The proposed method adaptively enables quantization either in the spatial domain or in the transform domain. The quantization in the spatial domain is directly conducted for residual signals without any transformations. For the areas for spatial quantization, bilateral filter is used to keep edge and not to generate artifact across boundaries. Experimental results show that the proposed method achieves coding gains of 3.6%, 4.0%, and 4.7% in BD-bitrate for random access, all-intra, and low-delay modes with the high efficiency configuration, respectively.