A novel 3D scanning technique for reflective metal surface based on HDR-like image from pseudo exposure image fusion method

Abstract

When scanning a metal workpiece with structured light, the reflected light can cause information loss or errors, ultimately affecting the reconstruction accuracy of the metal workpiece. Most studies use high dynamic range methods to overcome this issue, and a lot of research has been done on how to choose exposure time, but this kind of 3D scanning might lack efficiency. In this paper, we propose a new method. First, we calculate the relationship of the reference images and generate pseudo-exposure images through two groups of pictures, then we use image fusion to generate high dynamic images. Through the full factorial experimental design, the standard deviation of the reconstructed surface and the number of generated point clouds are recorded. By evaluating these indicators comprehensively, the best experimental parameters can be obtained. The experimental results show that the method proposed in this paper can effectively solve the issue of scanning reflective metal surfaces. Compared with other methods, the proposed pseudo-exposure method can reach the expected accuracy, the number of photographs taken can be reduced, and the scanning efficiency can be effectively improved.