摘要
MIHT(Modified Iterated Hough Transform)方法用于影像配准时,具有以直线段作为配准基元、线段之间不需同名关系等优点,同时亦存在着收敛速度慢、计算费时等不足。针对MIHT算法的缺点,提出将选权迭代最小二乘法引入到计算过程中,并且文中给出改进算法的具体实施流程。实验证明,与原算法相比较,改进算法的收敛速度和可靠性都有了明显的提高。
MIHT(Modified Iterated Hough Transform) is an image registration algorithm.It has advantages as straight-line primitives and none demands for correspondence between lines, while it also has a disadvantage as slow convergence.In order to overcome this shortcoming, an improved method that embraces MIHT and Huber Variable Weight Iteration Least Square Adjustment is brought forward.The workflow is introduced in detail.Moreover, experiments prove that the proposed registration method has obviously higher convergent speed and thus higher reliability for image registration.
出处
《测绘科学》
CSCD
北大核心
2008年第1期91-92,共2页
Science of Surveying and Mapping
关键词
线特征
MIHT
影像配准
选权迭代最小二乘法
line feature
MIHT
image registration
Huber Variable Weight Iteration Least Square Adjustment.
