A hybrid scheme incorporating stereo-matching and shape-from-shading for spatial object recognition

Abstract

Manufacturing flexibility can benefit from the use of non-contact computer-vision based shape-recovery methods. In this paper, a novel hybrid scheme is proposed, which combines stereo-vision and shape-from-shading recovery for obtaining more reliable absolute shape information about a viewed object. This iterative scheme first utilizes stereo-vision to obtain basic depth information along the boundary and through the surface of the object. Subsequently, the depth map is input to the shape-from-shading algorithm to obtain feature patterns. This information is fed back to the stereo-vision algorithm to improve upon the accuracy of the depth map. A cost function, in the form of a weighted sum of distances from smoothness and brightness, is utilized for the pattern matching. The performance of the proposed scheme was verified via experiments.