Three-dimensional reconstruction of highly reflective surfaces has always been an important and difficult problem in structured light technology. In this paper, a novel method by combining optimal exposure with the adjustment of the intensity of projection fringes by deep learning is proposed a single frame method is used to extract the overexposed region in the optimal exposure time series for composite projection. By using the statistical histogram of the number of overexposed pixels of the measured object under different projection conditions, and combining with the proportion of pixels in the image that are not exposed, the optimal exposure time series is calculated and the overexposed area is automatically extracted. The single-frame composite high dynamic image is obtained by adjusting the projection intensity through absolute coordinates, which improves the single-frame high dynamic range technique for highly reflective surfaces. Several simulations and experiments were conducted. The experimental results show that the proposed method can effectively improve the 3D reconstruction accuracy of the highly reflective surface, and the number of effective pixels in the point cloud can reach more than 98%.
|