基于改进Hausdorff距离的图像匹配快速算法

针对Hausdorff距离的特性及其在图像匹配中的应用,提出了一种基于改进Hausdorff距离准则的快速图像匹配算法。首先对图像进行小波分解和Harris角点检测后得到图像的特征点金字塔,然后利用改进的Hausdorff距离作为度量准则从特征点序列图像的最大尺度开始逐级向低尺度进行匹配。实验结果证明,该方法能在保证匹配精度的同时,有效加快匹配速度。

A Fast Image Registration Method Based on Advanced Hausdorff Distance

A new rapid image registration method was proposed based on Advanced Hausdorff Distance (AHD)，which is a criteria for two point sets.First，wavelet decomposition and Harris corner detector were used to construct the corner-point sequence.Then，the matching process began from the maximum level to the lower level of the sequence according to the similarity criteria of AHD.The experiment results showed that our method can reduce the running time while keeping high registration accuracy when the pyramid has multiple levels. The experiment result demonstrated the advantages of the method.