Adaptive Pattern Resolution for Structured Light 3D Camera System

3D depth measurement based on structured light has been gaining an increased attention, especially, in the field of robotics and automation. This is due to its capability of capturing a high accuracy of 3D point cloud in the trade-off with its scan speed. One way for a 3D camera to achieve a higher scan speed is combining a high frame rate of machine vision camera with an embedded projector. However, an embedded projector with Digital Micromirror Device (DMD) is often subject to a nonstandard pixel resolution. As such, to obtain a good quality of dense 3D point cloud, it is necessary to adapt structured light patterns to fit into the DMD pixel resolution at hand. This paper proposes a method of adaptive pattern generation that allows a maximum utilization of a non-standard resolution of DMD pixels for a structured light 3D camera. Experimental result shows that the method improves the density of a captured 3D point cloud by as much as 60%.

I. Introduction

Structured light 3D camera systems have been widely adopted to date, especially, in the fields of robotics [1]–[4], automation [5] [6] and industrial inspection [7]. Advantage of structured light 3D depth measurement systems includes a high precision and accuracy of depth measurement in the trade-off with a high-speed of depth measurement [8]. A structured light 3D camera system consists of one camera and one projector or of multiple cameras and a single projector [9]. A sequence of known patterns is projected on the scene and correspondence is made between the projected frame and the captured frame.