An improved adaptive rood pattern search algorithm for motion estimation

• Main Authors: Yu-Fang, Zheng, 鄭又方
• Other Authors: 鍾國亮
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/17071223114939943371

碩士 === 國立臺灣科技大學 === 資訊工程系 === 92 === Recently, Nie and Ma presented an efficient adaptive rood pattern search algorithm (ARPSA) for motion estimation. Experimental results reveal that their proposed ARPSA algorithm has better execution-time performance and better image quality when compared to the diamond search algorithm and the other algorithms. This short paper presents an improved ARPSA (IARPSA). Our proposed IARPSA presents a novel prediction strategy to narrow the search space required in the ARPSA and our proposed prediction strategy leads to better execution-time performance. Besides the computational advantage, our proposed prediction strategy also preserves the good image quality advantage as in the ARPSA. Under six real video sequences, experimental results demonstrate that our IARPSA has 18% average execution-time improvement ratio and 0.3dBs image quality improvement when compared to the ARPSA.