Single Pattern Structured Light Profilometry for In-Line Surface Inspection

Abstract

This paper introduces a novel scanning-based approach for dense surface reconstruction using optimization-based structured light profilometry (SLP) from a single projection pattern. Traditional methods, such as scanning laser-line triangulation and single-shot SLP, struggle to achieve results with both high density and accuracy at low acquisition rates. Increasing the acquisition rate typically requires either faster hardware or reduced relative scanning speed, imposing practical limitations. The presented method enables dense and accurate reconstruction of the passing sample from multiple images of a statically projected pattern by processing them together with the information about relative sample motion in an iterative optimization procedure that estimates the depth of the scene, the sample reflectivity and the ambient illumination. With the achieved accuracy of 19.9μm, this approach rivals high-resolution static measurements, offering applicability for industrial in-line surface inspection.