Enhancing perceived quality of compressed images and video with anisotropic diffusion and fuzzy filtering

Fuzzy filtering has recently been applied and optimized for reducing distortion in compressed images and video. In this paper, we present a method combining the powerful anisotropic diffusion equations with fuzzy filtering for removing blocking and ringing artifacts. Due to the directional nature of these artifacts, we have applied directional anisotropic diffusion. In order to improve the performance of the algorithm, we select the threshold parameter for the diffusion coefficient adaptively. Two different methods based on this approach are presented: one designed for still images and the other for YUV video sequences. For the video sequences, different filters are applied to luminance (Y) and chrominance (U,V) components. The performance of the proposed method has been compared against several other methods by using different objective quality metrics and a subjective comparison study. Both objective and subjective results on JPEG compressed images, as well as MJPEG and H.264/AVC compressed video, indicate that the proposed algorithms employing directional and spatial fuzzy filters achieve better artifact reduction than other methods. In particular, robust improvements with H.264/AVC video have been gained with several different content types.

► We present two effective algorithms for image and video artifact removal.
► The algorithms are applied on JPEG images, MJPEG and H.264/AVC video sequences.
► Subjective and objective quality assessment demonstrate the benefits our methods.