A Fast Method for Extraction of 3-D Information Using Multiple Stripes and Two Cameras

Abstract

This paper describes a method to extract 3-D information from two cameras in the scene on which multiple stripes are projected. Though a single camera cannot identify each stripe in a scene and its image when discontinuities of each stripe occur, wi th one more camera which our method employs the identification problern can be sol ved, because multiple stripes and two cameras give two constraints. One is the geometric constraint which gives the necessary condition for identification of each stripe. Another is the local constraint that features between images lie in constant order. After applying the geometric constraint, utilization of the local constraint enables identification of each stripe in a scene and its image. As a result, range data are obtained along multiple stripes. We also give a new method for computing camera parameters of 6 degrees of freedom which influence accuracy of 3-D information. They are derived mathematically by seeing the known cube.