A parallel motion estimation engine for H.264 encoding using the UMHexagonS algorithm

We present a fast motion estimation engine for the Unsymmetrical cross Multi-Hexagon-Grid-Search (UMHexagonS) algorithm used in H.264 video encoding. The architecture uses arrays of subtracting elements to maximize parallelism among search points on vertical or horizontal lines, and reduces sum of absolute differences (SAD) calculations. Performance is measured in terms of cycles required to encode Common Intermediate Format (CIF) video frames with 5 reference frames within a search range of 16. The number of cycles is reduced by around 99% compared with the JM 10.1 package running on a baseline Pentium 4 processor. The architecture thus achieves real time motion estimation at a clock speed of only 200MHz.

[1]  Liang-Gee Chen,et al.  Hardware architecture design of an H.264/AVC video codec , 2006, Asia and South Pacific Conference on Design Automation, 2006..

[2]  Liang-Gee Chen,et al.  Hardware architecture design for variable block size motion estimation in MPEG-4 AVC/JVT/ITU-T H.264 , 2003, Proceedings of the 2003 International Symposium on Circuits and Systems, 2003. ISCAS '03..

[3]  Chien-Min Ou,et al.  An efficient VLSI architecture for H.264 variable block size motion estimation , 2005, IEEE Transactions on Consumer Electronics.

[4]  Peter Pirsch,et al.  Array architectures for block matching algorithms , 1989 .

[5]  Choudhury A. Rahman,et al.  UMHexagonS algorithm based motion estimation architecture for H.264/AVC , 2005, Fifth International Workshop on System-on-Chip for Real-Time Applications (IWSOC'05).

[6]  Thomas Wiegand,et al.  Draft ITU-T recommendation and final draft international standard of joint video specification , 2003 .

[7]  Yong-Beom Cho,et al.  Hardware Architecture for Fast Motion Estimation in H.264/AVC Video Coding , 2006, IEICE Trans. Fundam. Electron. Commun. Comput. Sci..

[8]  Jianfeng Xu,et al.  Fast integer-pel and fractional-pel motion estimation for H.264/AVC , 2006, J. Vis. Commun. Image Represent..

[9]  Oscar C. Au,et al.  Predictive motion vector field adaptive search technique (PMVFAST): enhancing block-based motion estimation , 2000, IS&T/SPIE Electronic Imaging.

[10]  Baocai Yin,et al.  A novel parallel fast motion estimation algorithm , 2005, Proceedings of 2005 International Conference on Intelligent Sensing and Information Processing, 2005..

[11]  Kari Halonen,et al.  A hardware-based predictive motion estimation algorithm , 2005, 2005 IEEE International Symposium on Circuits and Systems.

[12]  Ilker Hamzaoglu,et al.  A high performance hardware architecture for an SAD reuse based hierarchical motion estimation algorithm for H.264 video coding , 2005, International Conference on Field Programmable Logic and Applications, 2005..

[13]  Michael Stegherr,et al.  Parameterizable VLSI architectures for the full-search block-matching algorithm , 1989 .

[14]  Oscar C. Au,et al.  Fast mode decision and motion estimation for H.264 (FMDME) , 2006, 2006 IEEE International Symposium on Circuits and Systems.

[15]  Yücel Altunbasak,et al.  SAD reuse in hierarchical motion estimation for the H.264 encoder , 2005, Proceedings. (ICASSP '05). IEEE International Conference on Acoustics, Speech, and Signal Processing, 2005..