基于改进ICP算法的地磁图匹配技术  

Geomagnetic-matching technology based on improved ICP algorithm

作　　者：任治新[1] 罗诗途[1] 吴美平[1] 刘颖[1] 

机构地区：[1]国防科学技术大学机电工程与自动化学院,长沙410073

出　　处：《计算机应用》2008年第B06期351-354,共4页journal of Computer Applications

基　　金：武器装备探索研究资助项目;高等学校博士学科点专项科研基金资助项目(20069998009);中国博士后科学基金资助项目(20070420186)

摘　　要：将地磁信息应用于惯性导航是组合导航技术研究的新方向。对地磁辅助导航的原理和发展现状进行了介绍,并在分析地磁特征的基础上,研究了利用地磁异常特征进行ICP匹配定位的方法。根据应用环境,对ICP算法进行了以下改进:首先,受模拟退火法的启发,提出采用扇形扫描法搜索最优匹配,推动ICP算法向全局最优迭代;其次,结合RANSAC法,有效剔除了地磁测量中的野值点,提高了算法的鲁棒性。仿真结果证实了改进算法的良好性能。Geomagnetism Aided Navigation （GAN） is a new branch of the integrated navigation techniques. The principle and the development of the GAN were introduced and the features of the geomagnetism field were analyzed, herative Closest Point （ICP） matching method based on geomagnetic anomaly map was studied, The traditional ICP algorithm was improved in two aspects： firstly, inspired by simulated annealing, sector scan method was proposed to search the optimal match. The sector scan method could help ICP jump out of the local minimum convergence field and to the global one, improving the precision of localization greatly. Secondly, the RANSAN method was used to remove the outliers, which effectively enhanced the robustness of the matching method. Computer simulation was used to test the proposed method and good performance was received.

关 键 词：迭代最近点算法 地磁匹配 随机抽样一致性算法 鲁棒估计 地磁辅助导航 

分 类 号：TP391.7[自动化与计算机技术—计算机应用技术]