3D model generation of complex objects from multiple range images

This paper demonstrates on three dimensional (3D) model generation of complex objects from multiple range images, which is usually utilized in reverse engineering and computer vision. The object covers a series of key techniques. Registration of different range image is one of the most important algorithms. In this paper, we present an efficient and uniform approach to consolidate all unorganized point clouds into one reference frame, which is not similar to the traditional registration algorithm, Iterative Closest Point (ICP) algorithm. A planar grid, engraved by cross lines, is used to define the reference frame where all the unorganized points will be transformed. The intersection points of all cross lines are detected from each range image, and their correspondent points in the common reference frame are also obtained, therefore, all the transformation parameters will be calculated with least squares adjustments, and all the unorganized points will be transformed into the common reference frame to make a complete surface.