對於多尺寸方塊的運動向量估計，本論文提出一個有效率的以合併為基礎的演算法，用來降低H.264視訊編碼系統中運動向量估計步驟的計算複雜度。 首先對44尺寸的小方塊用完全搜尋的方式作整數像素精確度的運動向量估計，並且為每一個44小方塊保留一些運動向量做候補。這些候補的運動向量被用來決定相鄰方塊是否可以合併成為H.264所定義的其他形狀的方塊。因為只對44小方塊作運動向量估計的程序，所以其它尺寸方塊的運動測量計算負載將被節省下來。對於在外部 （inter）和內部 （intra）模式之間資料量－失真間 （RD）的最佳模式選擇，門檻值（threshold）決策方法將被採用以節省外部 （inter）- 168， 816， 88模式和內部模式的計算負載。模擬結果顯示出此簡單演算法的效能接近於 － 當1616， 168， 816， 88， 84， 48和44尺寸方塊皆被啟用時，H.264視訊編碼系統的測試程式JM7.3，而計算複雜度則明顯地降低。

　　This Thesis presents an efficient merge-based algorithm for variable-size block matching motion estimation to reduce the computation load of motion estimation process in H.264 video encoder. Initially, fixed-size 44 small blocks are used for the integer-pixel accuracy full-search motion estimation and several candidate motion vectors are reserved for each 44 blocks. These motion vectors are determined further that if the neighboring blocks could be merge into other predefined block types or not. Because the motion search is performed only at the smallest block size 44, the computational load of other block size motion estimation can be saved. While in RD optimized mode decision between inter and intra mode, the threshold decision method is adopt to save more computation load for inter - 168, 816, 88 and intra mode. The simulation results show that the performance of this simple algorithm is close to H.264 video encoder JM7.3 with all block modes -1616, 168, 816, 88, 84, 48, and 44 are enabled while the computational complexity is significantly reduced.

REFERENCE

[1] A. Puri, H. Hang, and D. Schilling, “Interframe coding with variable block size motion compensation,” in Proc. GLOBECOM'87, pp. 65–69.
[2] D. Marr and S. Ullman, “Directional selectivity and its use in early visual processing,” Proc. Royal Soc. London, vol. 211.B, pp. 151–180, 1981.
[3] G. Sullivan and T. Wiegand, “Rate-distortion optimization for video compression,” IEEE Signal Processing Magazine, Vol.15, pp. 74-90, Nov. 1998.
[4] H. Jelveh and A. Nandi, “Improved variable size block matching motioncompensation for video conferencing applications,” in Digital SignalProcessing, A. Cappellini and A. Constantinides, Eds. Berlin, Germany: Springer-Verlag, 1991, pp. 391–396.
[5] I. Rhee, G. Martin, S. Muthukrishnan, and R. Packwood, “Quadtree-Structured Variable-Size Block-Matching Motion Estimation with Minimal Error,” IEEE Trans. Circuirs Syst. Video Technol., vol. 10, pp. 42-50, Feb. 2000.
[6] ITU Telecom. Standardization Sector of ITU, “Video Coding for Low Bitrate Communication,” Draft ITU-T Recommendation H.263 Version 2, January 1998.
[7] ITU-T. 2001. “H.26L Test Model Long Term Number 8(TML-8) draft0.” International Telecommunication Union, July 2001.
[8] ITU-T Rec. H.264 / ISO / IEC 11496-10, “Advanced Video Coding”, Final Committee Draft, Document JVT-E022, September 2002
[9] M. Chan Y. Yu, and A. Constantinides, “Variable size block matching motion compensation with applications to video coding,” Proc. Inst. Elect. Eng., pt I, vol. 137, no. 4, pp. 205-212, Aug. 1990.
[10] MPEG Video Group, “MPEG-4 video verification model version 8.0,” ISO/IECJTC1/SC29/WG11n1796, July 1997.
[11] Woong Il Choi, Jeyun Lee, Sungmo Yang, Byeungwoo Jeon, “Fast motion estimation and mode decision with variable motion block sizes”, Proc. SPIE VCIP, July 2003.
[12] Y.K. Tu, J.F. Yang, Y.N. Shen, and M.T. Sun, “Fast Variable-Size Block Motion Estimation Using Merging Procedure with an Adaptive Threshold,” IEEE ICME 2003.