一种频域亚像元影像匹配方法

A Subpixel Matching Method in Frequency Domain

频域相位匹配方法可以分为空域Dirac Delta函数法和频域直接求解法两类,前者一般只能达到整数像元匹配精度,而后者可实现亚像元匹配精度,但算法复杂度高,匹配可靠性较前者低。基于其各自的特点,文章将上述两类方法有机结合,提出了一种可靠、高精度的优化相位匹配法。通过对不同的混叠量、信噪比影像进行匹配实验,验证了文章提出的匹配方法对混叠具有一定的抑制能力,对噪声有较强的鲁棒性。对于纹理丰富、存在少量混叠的影像匹配误差优于1/100个像元。

There are two kinds of the phase matching method including the Dirac Delta method and the frequency domain method. The former matching accuracy can achieve only pixel level generally, and the latter can achieve subpixel level but being more complicated and less reliable than the former. This paper presents an optimized phase matching method by integrating them. This method was tested with some kinds of image. Experimental data presented here show that the new method gives superior matching precision（1/100 pixel） for an image pair with rich textures and a little aliasing.