图象分块平均绝对差匹配并行算法

Image Blocking Matching Parallel Algorithms on Mean Absolute Difference

本文针对图象匹配算法中的乘法运算与差运算的一类并行计算问题 ,设计了一种图象分块平均绝对差匹配并行算法。该算法对于 N× N的参考图像、与处理元阵列大小相同的 M× M的实时图象 (N=BK,K=M,B>1) ,可通过将参考图像分成 B× B个图像块的分块匹配来实现 。

In accordance with the computer demand of multiplication and difference in image matching algorithms . The image blocking matching parallel algorithms on mean absolute difference are designed. The algorithms realizes image matching by dividing the reference image into B×B image blocks with N by N the reference image, M by M real time image same size as processing unit array(N=BK, K=M, B>1), and has a higher parallel efficiency.