快速运动估计中一种改进的块匹配免疫算法

An improved block-matching immune algorithm in fast motion estimation

图像的运动估计中,基于块匹配的免疫算法由于算法的随机性,使得抗体群中不匹配块增多,导致了免疫算法运算量增大。改进算法搜索过程中,通过快速识别出搜索窗内不可能成为匹配块的候选块,并把这些块消除掉,不对其进行匹配误差运算,从而可以大幅度减少免疫算法运算量。同时,由于改进算法使得候选块的匹配可能性提高,使得相对于传统块匹配免疫算法,重建图像的均峰信噪比进一步提高,进而重建图像质量得以提高。仿真实验结果验证了改进算法的上述优点。

Because of the randomicity of Block match Immune Algorithm （BIA）, the number of mismatch block and computation load increase considerably. In the new algorithm, those blocks of being unable to match will be picked out and abolished quickly, so it does not need match error calculating. That means the computation load decreases greatly. At the same time, the selected blocks have more match possibility and results in the more high Peak Signal-to-Noise Ratio （PSNR） and better rebuilt image. At last, the computer simulations verified the advantages mentioned above.