摘要
相对定向的解算需要利用足够数量的同名像点。针对同名像点对应的空间点分布处于临界配置时相对定向不能获得唯一解的问题,该文对摄影测量和计算机视觉领域内常用的相对定向算法进行了研究。分析了它们的联系和区别,利用物方形式的共面条件方程推导了临界配置中危险表面的通用方程,分析了直线和平面配置下相对定向的退化情况,并利用奇异值分解推导了不同相对定向算法的临界配置条件。
The determination of relative orientation needs sufficient homologous image points, but we can＇t get the unique solution of relative orientation if the spatial points corresponding to these homologous image points set in the critical configuration. In this paper, the relative orientation algorithm used within the field of photogrammetry and computer vision was studied, and their relationship and distinction was analyzed. The general equation of dangerous surface in the critical configuration was deduced using the coplanar condition function in the object space form, the critical configurations of the straight line and plane were analyzed, and the critical conditions of different relative orientation algorithms was computed using singular value decomposition.
出处
《测绘科学》
CSCD
北大核心
2016年第10期19-23,共5页
Science of Surveying and Mapping
基金
国家"973"计划项目(2012CB720000)
关键词
相对定向
临界配置
共面条件方程
奇异值分解
relative orientation
critical configuration
coplanar condition function
singular value decomposition
