测绘学报 ›› 2019, Vol. 48 ›› Issue (3): 322-329.doi: 10.11947/j.AGCS.2019.20170672

• 摄影测量学与遥感 • 上一篇    下一篇

大角度立体像对相对定向的混合共轭梯度算法

李佳田1,2, 王聪聪1,2, 贾成林1,2, 牛一如1,2, 王瑜1,2, 张文靖1,2, 吴华静1,2, 李键1,2   

  1. 1. 昆明理工大学国土资源工程学院, 云南 昆明 650093;
    2. 昆明理工大学云南省高校高原山区空间信息测绘技术应用工程研究中心, 云南 昆明 650093
  • 收稿日期:2017-11-27 修回日期:2018-10-20 出版日期:2019-03-20 发布日期:2019-04-10
  • 通讯作者: 王聪聪 E-mail:1083719493@qq.com
  • 作者简介:李佳田(1975-),男,博士,教授,研究方向为数值最优化方法与机器场景理解。E-mail:ljtwcx@163.com
  • 基金资助:
    国家自然科学基金(41561082;41161061)

A hybrid conjugate gradient algorithm for solving relative orientation of big rotation angle stereo pair

LI Jiatian1,2, WANG Congcong1,2, JIA Chenglin1,2, NIU Yiru1,2, WANG Yu1,2, ZHANG Wenjing1,2, WU Huajing1,2, LI Jian1,2   

  1. 1. Faculty of Land Resource Engineering, Kunming University of Science and Technology, Kunming 650093, China;
    2. Surveying and Mapping Geo-Informatics Technology Research Center on Plateau Mountains of Yunnan Higher Education of Kunming University of Science and Technology, Kunming 650093, China
  • Received:2017-11-27 Revised:2018-10-20 Online:2019-03-20 Published:2019-04-10
  • Supported by:
    The National Natural Science Foundation of China (Nos. 41561082;41161061)

摘要: 无初值依赖的快速收敛是大角度相对定向解算的关键所在。为此,本文提出一种混合共轭梯度算法,具体过程是:① 采用随机爬山算法对给定的相对定向元素初值进行随机扰动,产生保证优化方向的初值;② 局部优化中以超线性收敛的共轭梯度法取代相对定向中的最速下降法,以提高其收敛速度;③ 全局收敛条件为计算误差小于规定的限差。对比试验表明,混合共轭梯度算法无初值依赖性,具有较高的解算精度和较少的迭代次数。

关键词: 相对定向, 大角度, 全局收敛, 随机爬山算法, 共轭梯度法

Abstract: The fast convergence without initial value dependence is the key of large angle relative directional solution. Therefore, a hybrid conjugate gradient algorithm is proposed in this paper. The concrete process is:① stochastic hill climbing(SHC) algorithm is used to make random disturbance to the given initial value of the relative directional element, and the new value to guarantee the optimization direction is generated; ② In local optimization, super-linear convergent conjugate gradient method is used to replace the steepest descent method in relative orientation to improve its convergence rate; ③ The global convergence condition is that the calculation error is less than the prescribed limit error. The comparison experiment shows that the method proposed in this paper is independent of initial value, has higher accuracy and fewer iterations.

Key words: relative orientation, big rotation angle, global convergence, stochastic hill climbing algorithm, conjugate gradient algorithm

中图分类号: