Fringe Image Enhancement for Structured Light 3-D Measurement of Low-Reflective Objects

3-D measurement of low-reflective objects presents significant challenges in industrial inspection and diagnostics. This article introduces a new fringe image enhancement method for structured light 3-D measurement, tailored specifically for low-reflective objects. A single-channel Transformer Retinex (SCTR) enhancement network is designed for processing fringe images. The SCTR network’s framework integrates the Retinex model with Transformer blocks and incorporates a novel modulation loss function tailored for the preservation and enhancement of fringe texture details. Since the network training process requires extensive paired datasets, we have created a digital twin of the structured light system to produce a large virtual dataset, supplemented by real datasets for deep learning. With the developed stereo structured light system, accurate 3-D measurement of low-reflective objects was accomplished. Experimental results demonstrated that the proposed method outperforms state-of-the-art techniques in both image quality enhancement and 3-D measurement accuracy, providing an efficient solution for measuring low-reflective objects.