中文摘要：

为克服传统的相似性度量容易受到噪声、遮挡和成像机理等因素影响的缺点,结合人的认知过程,提出了一种分层的模板匹配算法。首先利用了统计指标来对候选匹配区域进行预标记,其次通过对Hausdorff相似性度量的改进来提高其对遮挡、异源图像匹配的鲁棒性。实验结果证明了该方法能够有效地减少搜索区域大小,提高了遮挡情况下的匹配精度,验证了算法的有效性。

英文摘要：

To deal with the defect that traditional matching methods were susceptible to noise,occlusion and change of imaging mechanism,this paper presented a hierarchical template matching algorithm,according to human cognitive process. Firstly it cut candidate matching areas by using the statistical index,then improved the robustness though an improvement algorithm based on MHD. The experimental results have impressively indicated the effectiveness of the proposed method to keep candidate matching areas down and improve the robustness.