Accuracy Improvement of Fringe Projection 3D Scanner Using Photometric Stereo Method

Abstract

Fringe projection (FP) technique uses a fringe pattern in a determined direction to measure the threedimensional shape of the object. This paper surveys the effects of the object surface geometry on the accuracy of the three-dimensional reconstruction. Inter-reflection in concave areas, a significant change of phase in step edges, and lower signal to noise ratio in slope areas are the object geometry effects on the phase retrieval process. We proposed an FP technique that uses a photometric stereo to classify the image pixels based on the geometric properties of the object surface for selection of appropriate pattern direction. Experiments demonstrate that our low-cost integrated system can improve the measurement's accuracy of the objects with complex geometry from 58 microns to 28 microns.