DIFFERENTIAL FRAME ANALYSIS AND SELECTIVE MOTION ESTIMATION APPROACH FOR AN INTER-FRAME JPEG2000 VIDEO CODING SYSTEM

Low-complexity motion estimation algorithms restricting block matching to certain blocks containing high motion are proposed in the context of a JPEG2000 based video coding system. The blocks subjected to motion estimation / motion compensation are identified based on image analysis of differential frames where simple arithmetic indicators are employed to determine motion content. It is shown that the block-selective approach achieves the PSNR quality of a full motion estimation scheme with about 60% of the blocks for many test sequences and delivers similar rate distortion performance compared to three step search if complexity is fixed at an equal level.