A ROBUST METHOD FOR REGISTRATION AND SEGMENTATION OF MULTIPLE RANGE IMAGES

Registration and segmentation of multiple range images are important p roblems in range image analysis. We propose a new algorithm of range d ata registration and segmentation that is robust in the presence of ou tlying points (outliers) like noise and occlusion. The registration al gorithm determines rigid motion parameters from a pair of range images . Our method is an integration of the iterative closest point (ICP) al gorithm with random sampling and least median of squares (LMS or LMedS ) estimator. The segmentation method classifies the input data points into four categories comprising inliers and 3 types of outliers. Final ly, we integrate the inliers obtained from multiple range images to co nstruct a data set representing an entire object. We have experimented with our method both on synthetic range images and on real range imag es taken by two kinds of range finders. The proposed method does not n eed preliminary processes such as smoothing or trimming of isolated po ints because of its robustness. It also offers the advantage of reduci ng the computational cost. (C) 1995 Academic Press, Inc.