摘要
本文提出一种根据刚体运动前后生成的２Ｄ点对应数据集合估计其运动参数的方法，考虑到集合中存在部分匹配错误的对应点，文中利用最小截平方和估计和递归残差分析的方法来剔除这些出格数据，从而获得刚体空间运动的正确参数，并且不需要任何先验知识．实验结果表明了算法的有效性．
Given the 2 D point correspondence set of a scene containing moving rigid object, this article proposes a method to estimate its motion parameters. Because there exists several incorrect correspondence in the set, the least trimmed square estimator and recursive residuals analysis is adopted to deal with the outliers and get the correct motion parameters. It does not need any prior knowledge. Computer simulation results demonstrate the effectiveness and efficiency of the algorithm.
出处
《信息与控制》
CSCD
北大核心
1999年第5期339-344,共6页
Information and Control
关键词
刚体运动
参数估计
遗传算法
鲁棒性
rigid motion, parameter estimation, least trimmed squares, genetic algorithm, robust, outliers, recursive residuals
