A Hybrid Direction-Based Block-Matching Algorithm for Fast Motion Estimation in Video Coding
Block motion estimation is one of the most important modules in the design of any video coder. It is widely used in many motion compensated video-coding standards like MPEG and ITU-T for multimedia and video/ teleconferencing applications. As Motion Estimation is the most time consuming module, the existing Block Matching Algorithms provide a trade-off between the time taken for compression and the quality achieved. In this paper, we propose a hybrid Direction-based Block Matching algorithm with efficient entropy encoding technique to minimize the search time on block matching and relatively high speed encoding method to achieve higher compression ratio. The improved version of Direction-based Block Matching reduces the search time by utilizing Cross-Diamond Search (CDS) and Efficient Three-step Search (ETS) rather than the conventional full search algorithm. Experimental results show that the proposed algorithm provides a faster search with minimum distortion when compared to the conventional CDS and ETS.
Keywords: Video Coding, Motion Estimation, Entropy Encoding, Multimedia
Mr Ezhilarasan Murugesan
Pondicherry Engineering College
Professor and Head, Computer Science & Engineering and Information Technology, Pondicherry Engineering College
Information Technology, Pondicherry Engineering College,Pondicherry, India. He received the M.E. from
College of Engineering, Guindy, Chennai and Ph.D. in computer science from Alagappa University, Karaikudi, India in 1984 and 1995 respectively. His research interests include natural language processing and data compression. He is a life member of Computer Society of India and Indian Society for Technical Education.