Direction-Adaptive Motion Estimation (DAME) for Efficient Video Compression

Motion estimation (ME) is widely accepted as an effective tool in video coding to reduce temporal correlation between frames and yields significant improvement in compression gain while sustaining the picture quality. Among various ME schemes, block-matching motion estimation (BMME) is the most popular approach due to its simplicity and efficiency. The BMME algorithms use search patterns to find out the best matched block. However, a single search pattern does not prove efficient in finding best matched block for all motion types. In this research, an efficient direction-adaptive motion estimation (DAME) algorithm is proposed which adaptively selects shape and size of the patterns based on motion content. Performance analysis of DAME algorithm on the reference code JM of H.264/AVC platform reveals that the proposed algorithm outperforms other existing BMME algorithms in search points without degrading the picture quality.