顾及形变的影像边缘ICP匹配技术

ICP image matching technology considering distorted image edge

匹配技术是遥感影像处理中的重要基础技术,影像在辐射、几何信息方面的差异,将会严重干扰影像间的相似特征匹配。本文提出了针对大差异影像的最邻近点迭代匹配算法(ICPDI)。首先改进Canny算子的参数设置方式,以便选择影像的长边缘作为影像匹配的显著特征;然后以线特征之间的距离作为影像相似性的评估标准,达到利用全局相似特征抑制局部差异的目的;最后给影像边缘附加形变函数,从而扩展ICPDI匹配技术的适用范围。通过不同差异类型遥感影像间的匹配试验,验证了本文方法的精度和适应性。

Matching technology is an important basic technology in remote sensing image processing.The difference between radiation and geometric information of images seriously interferes with similar feature matching between images.This paper proposes an algorithm of ICPDI for large difference images.Firstly,it improves the parameter setting method of the Canny operator for selecting the long edge of the image as a prominent feature of image matching.Then,it takes the distance between line features as the similarity evaluation standard for achieving the purpose of suppressing local differences by using global similar features.Finally,it adds a deformation function to the edge of the image for extending the scope of ICPDI matching technology.The accuracy and adaptability of the proposed method are verified by matching experiments between different types of remote sensing images.