低空无人机影像自动相对定向  

Automatic Relative Orientation Method of Low-Altitude UAV Image

作　　者：郭岚[1] 陈明杰 GUO Lan;CHEN Mingjie(College of Geomatics, Xi’an University of Science and Technology, Xi’an Shaanxi 710054, China)

机构地区：[1]西安科技大学,陕西西安710054

出　　处：《北京测绘》2018年第12期1483-1488,共6页Beijing Surveying and Mapping

基　　金：国家自然科学基金青年科学基金项目(41501571)

摘　　要：针对传统无人机影像相对定向方法迭代不收敛和自动化程度较低的问题,本文提出了一种自动化程度更高且能有效收敛的方法。首先使用SIFT算法自动提取特征点,通过RANSAC算法剔除错误匹配的点并迭代计算基本矩阵,然后在理论上推导了基本矩阵及本质矩阵和相对定向元素的关系,从而导出相对定向元素的解,以此为初始值,利用严格相对定向模型迭代求解相对定向元素的值,直到其结果收敛为止。最后经过实验对比分析了小倾角和大倾角的无人机影像相对定向结果,实验证明了本文方法的有效性和可用性。In view of the problem that the traditional UAV image relative orientation method is not converge and the degree of automation is low,this paper proposes an automaticandconvergent method.At first,we automatically extract feature points by using scale-invariant feature transformalgorithm,thenwe match key points and eliminates the mismatched points and iteratively calculates the fundamental matrix by using RANSAC.Secondly,in this paper,the relationship between the fundamental matrix and the essential matrix and the relative orientation elements is theoretically derived.Thereby,the result of the relative orientation elements are derived.Taking the result as initial value,we use the strict relative orientation model to iteratively calculate the value of the relative orientation elements until result converges.Finally,we make a set of comparative experiments about large dip and small dipUAV image and get relative orientation elementfrom fundamental matrix.Taking the result as the initial value for traditional relative orientation model,we get a more accurate result.The experiments results show that the method of this paper is effective and usable.

关 键 词：尺度不变特征变换 无人机(UAV) 相对定向 基本矩阵 本质矩阵 

分 类 号：P231[天文地球—摄影测量与遥感]