由2D点匹配恢复刚体运动参数的鲁棒估计

ROBUST ESTIMATION OF MOTION PARAMETERS OF A RIGID BODY WITH 2D POINT MATCHES

本文提出了一种基于最小平方中值原理的运动参数估计的鲁棒方法，能够有效地克服由于出格点所产生的误差，并且不需要任何参数初值．大量模拟实验显示，本文方法优于Ｈａｒａｌｉｃｋ等人的Ｍ－估计方法［７］．

This correspondence proposes a robust estimation method based on least-square median principle for motion parameters,which works well in case where exist outliers without any prior knowledge.Simulation shows that the algorithm is better than M-Estimation based algorthm proposed by Harelike et al[7].