3D reconstruction of a strongly reflective surface based on binocular line-structured light

This paper proposes a 3D reconstruction method utilizing binocular line-structured light to address the accuracy errors caused by strong light reflection on the surface of objects in industrial production. This method involves an algorithm that employs image transformation, threshold segmentation, and image fusion to extract the center of the laser stripe from multiple images. Two monocular systems, each consisting of a light plane and a single camera, as well as a binocular system, were established. These systems generated three sets of point cloud data, which were then integrated using point cloud integration technology to obtain a set of high dynamic range 3D data. Experimental results showed that within a measurement range of 600 mm by 800 mm and at a measurement distance of 1200 mm, the root mean square (RMS) value was calculated to be 0.041 mm, with an overall error spread of ±0.07mm (0.13%). The point cloud data in the experimental scenarios presented in this paper demonstrated a maximum improvement of 121% compared to the traditional methods.