|
|
|
| A Fast Multi-Pattern Motion Estimation Algorithm Based on Quantum Clonal Selection |
| Liu Zhen① Bai Zhong-ying① Shi Jin-ming② Lin Zhao-wen① |
①(School of Computer Science, Beijing University of Posts and Telecommunications, Beijing 100876, China)
②(National Satellite and Meteorologic Centre, Beijing 100081, China) |
|
|
|
|
Abstract A novel fast multi-mode motion estimation algorithm based on the quantum clonal selection is proposed. According to the predictive motion vectors derived from spatio-temporal coherence of video sequences, the type of searching modes can be confirmed adaptively. In the stationary macroblock mode, current search is terminated immediately. In the mild motion mode, CDS method is used directly. In the drastic motion mode, a combined QCS/CDS search strategy is utilized. Firstly, the QCS method is used to search the appropriate optimal motion vector, then the CDS method is used to search global optimal motion vector. Experimental results show that the algorithm can speed up the process of motion estimation with fewer search points, while maintaining the average PSNR by using the full search algorithm with little loss.
|
|
Received: 23 October 2007
|
|
|
|
Corresponding Authors:
Liu Zhen
|
|
|
|
|
|
|
2008, Vol. 30 
