Abstract
A model-based positioning method for 3 D objects in stereo images is proposed in this paper. In this method, first a pair of edge enhanced images are generated from an original stereo pair, and rough range data are calculated from them. Next, a 3 D model is rotated and translated by using a set of space transformation parameters to make a projected image. Edges in the projected image and the transformed 3 D model are compared with the left edge enhanced image and the range data respectively to evaluate the space transformation parameters. In order to decrease the total number of calculations for the matching, Genetic Algorithm; GA is employed. A set of parameters which gives the best matching ratio is determined quickly according to GA.
