A post-processor for reducing temporal busyness in low-bit-rate video applications

摘要

This paper presents an original post-processing algorithm to reduce the so-called temporal busyness phenomenon in low-bit-rate video coded sequences. The post-processor is designed to be implemented in digital video receivers; in this case, only reconstructed video is available with no reference to coding parameters. The proposed technique first identifies the distortion in decoded video, then reduces both mosquito noise and discrete cosine transform (DCT) basis images affect by appropriate 3D (2D+t) adaptive filtering, while preserving the sharpness and naturalness of the reconstructed video signals. The mosquito noise is reduced by 2D signal-adaptive filtering, and DCT basis image effects are reduced by temporal DCT median filtering. Experimental results show that the proposed post-filtering process improves the visual quality of video significantly.