A Cost-Effective Single-Shot Structured Light System for 3D Shape Measurement

Abstract

Single-shot three-dimensional (3D) shape measurement techniques have attracted extensive researches, as they are suitable for dynamic measurement. In this study, a cost-effective single-shot structured light system (CES-SLS) is proposed for 3D shape measurement by using a color camera and a normal projector. With the aid of two planar mirrors and an isosceles right-angle mirror, a single color camera functions as a binocular vision. A single-shot color random speckle pattern (CRSP) is used to encode the object by the normal projector. The dense corresponding points (DCP) are derived by the spatial-temporal correlation of RGB image subsets instead of the spatial correlation of gray image subsets in the conventional method. An inverse compositional Gaussian Newton (IC-GN) iteration method with the second-order shape function is then introduced to find the sub-pixel corresponding points (CP). Furthermore, a geometrical 3D recovery method is presented to calculate the 3D point by minimizing the re-projection error. The experimental results demonstrate the comparative advantages of the proposed CES-SLS against the system using a single-shot random speckle pattern (RSP) and the curve fitting sub-pixel method in the aspects of measurement accuracy and noise robustness. In addition, the re-projection error of each 3D point from the geometrical method is smaller than that of the conventional mid-point.