面向组特征的航空影像直线匹配方法

Linear Matching Method of Aerial Imagery for Group Features

针对直线匹配结果存在"一对多""多对一""多对多"的问题,提出一种基于组特征的航空影像直线匹配方法。该方法在直线提取结果的基础上,首先通过区域和角度双重约束对左、右影像上提取的直线分别进行编组,然后采用核线约束、单应矩阵约束、相似性约束对组直线特征进行匹配及分裂,最终确定同名直线。采用某城区的数字航空影像进行直线匹配实验,取得可靠的匹配结果,验证该方法的有效性及可行性。

Aiming at solving the nonuniqueness problem in the process of stereo matching,an algorithm of linear matching of aerial imagery based on group features is proposed.First,it groups the straight lines extracted from left and right images with area and angle constraints respectively.Then,it matches and breaks the group features by epipolar,homography and similarity constraint.This article uses the digital aerial images of a certain city for straight line matching based on the group characteristics.The results indicate that this method works well and thus demonstrate its validity and feasibility.