Generic design methodology for the development of three-dimensional structured-light sensory systems for measuring complex objects

Abstract

Structured-light (SL) techniques are emerging as popular noncontact approaches for obtaining three-dimensional (3-D) measurements of complex objects for real-time applications in manufacturing, bioengineering, and robotics. The performance of SL systems is determined by the emitting (i.e., projector) and capturing (i.e., camera) hardware components and the triangulation configuration between them and an object of interest. A generic design methodology is presented to determine optimal triangulation configurations for SL systems. These optimal configurations are determined with respect to a set of performance metrics: (1) minimizing the 3-D reconstruction errors, (2) maximizing the pixel-to-pixel correspondence between the projector and camera, and (3) maximizing the dispersion of the measured 3-D points within a measurement volume, while satisfying design constraints based on hardware and user-defined specifications. The proposed methodology utilizes a 3-D geometric triangulation model based on ray-tracing geometry and pin-hole models for the projector and camera. Using the methodology, a set of optimal system configurations can be determined for a given set of hardware components. The design methodology was applied to a real-time SL system for surface profiling of complex objects. Experiments were conducted with an optimal sensor configuration and its performance verified with respect to a nonoptimal hardware configuration.