Registration And Integration Methods Of Multiple Range Images

There is growing interest in automatic object model construction via view registration and integration for applications ranging from reverse CAD engineering PR, visual museum to fashion design and entertainment industries. In this paper, the problem of registration and integration of multiple range images is addressed. Multiple range images can be obtained by turning the 3D object different angles around sensor. The range images of different viewpoints are registered to estimate the relative transformations among the viewpoints, and then integrated to obtain a complete 3D geometric description of object. Firstly, we introduce the principal of three dimensional sensors and our data collecting system, and two kinds of data errors in the range images. Next we propose a ICR-based registration algorithm for two partially overlapping range images. We consider all main components of ICP algorithm, such as control point selection~ the nearest point search.. corresponding pair抯 weighting~ the valid determination of correspondence and so on. Last we propose a two phase integration method. In the first phase, we use a volumetric method to get the integrated point set, which scale is varied and is easily on control. In the second phase, from the fused point set, we reconstruct the surface model. We calculate signed distance from every point in R3 to the approximated surface and extract an iso-surface from the distance field by an marching cubes algorithm. Experiments demonstrate that the method can satisfy the requirements of computing cost and accuracy of object model reconstruction. Our registration and integration methods provide an easy way to fast automatic model recovery.