3D-IN LOOP FILTER FOR VIDEO COMPRESSION BASED ON MOTION COMPENSATED TEMPORAL PIXEL TRAJECTORIES

Abstract

For H.264 standardization, the use of a deblocking filter was introduced in order to improve the quality of video compression through blocking artifacts reduction. Subsequently, the in-loop filters such as deblocking filter, sample adaptive offset, and adaptive loop filter are launched for the present HEVC standardization. However, both filters use spatial data from a frame of the video sequence only despite the fact of temporal correlation within frames. Hence, the quality of filtering process is restricted. Therefore to conquer this problem, we can apply 3Dfilter; the three dimensions being the two spatial coordinate and the time. The intention of this paper is to explore the effectiveness of a new in-loop filter that combines both spatial and temporal information, through quality improvement. The filter only requires a small overhead while using the temporal information transmitted in the bit stream to rebuild the individual motion trajectory of every pixel in a frame at the codec. This temporal information is then accustomed to performing pixel-wise motion compensated temporal filtering. The paper describes a nonlinear temporal filtering algorithm using motion estimation and compensation for reducing noise in video sequences. The background of the filtering technique concept is described and simulation results on several video sequences are presented. The new filter has been integrated with deblocking filter and it is revealed that the filter achieves better performance than the state-of-the-art codec H.264/AVC, which uses a deblocking filter as in-loop filtering over a large range of sequences and bit rates.