全部像元参加的影像几何匹配算法

发布时间：2019-05-23 01:31

【摘要】：影像几何匹配之目的是从确定对应的影像中提取其间的几何转换参数,在摄影测量测图、变形物面监测和飞行平台动态性能分析等方面有许多实用需求。不同于普遍悉知的特征点影像匹配算法,本文研究的算法是把对应影像内的全部像元素参加匹配运算,从而按最小二乘准则求解对应影像间的几何转换参数值。文中阐述了算法原理,包括灰度对应方程、信息量不等式及最小二乘求解方法,尤其是详细探讨了灰度信息的小波分析及由信噪比计算信息量的方法。为验证本文的理论和方法,采用了直升机载视频相机的序列影像进行了试验,给出了相对定向元素和视差格网的两种解算结果。最后给出了本文方法与传统特征点匹配方法可利用信息量不等式进行对比分析的结论。
[Abstract]:The purpose of image geometric matching is to extract the geometric conversion parameters from the corresponding images, and there are many practical requirements in photogrammetry, deformation surface monitoring and flight platform dynamic performance analysis. Different from the well-known feature point image matching algorithm, the algorithm studied in this paper is to take part in the matching operation of all the image elements in the corresponding image, so as to solve the geometric conversion parameters between the corresponding images according to the least square criterion. In this paper, the principle of the algorithm is described, including gray correspondence equation, information inequality and least square solution method, especially the wavelet analysis of gray information and the method of calculating information from SNR are discussed in detail. In order to verify the theory and method of this paper, the sequence image of helicopter-borne video camera is used to carry out the experiment, and two kinds of calculation results of relative orientation element and parallax grid are given. Finally, the conclusion that the information inequality can be used to compare the proposed method with the traditional feature point matching method is given.
【作者单位】：武汉大学遥感信息工程学院;中国测绘科学研究院;
【基金】：国家自然科学基金(41371425)~~
【分类号】：TP751

【相似文献】