Error Analysis and Reduction in 3D Reconstruction Based on Structured Light

In the process of structured light 3D reconstruction, the nonlinear response of projection equipment will greatly affect the accuracy of reconstruction. This paper introduces the reconstruction principle and process steps of n-step phase shift method with, complementary Gray code in, monocular inverse camera method, and analyzes the influence of phase shift error and phase error on the reconstruction results under different phase shift steps. In order to solve the contradiction between speed and accuracy of 3D reconstruction, this paper proposes to use multi-layer neural network and gray neighborhood information to solve the wrapped phase. Experimental results show that the proposed method can effectively compensate the reconstruction error, and can achieve a reconstruction accuracy of uncompensated 24 steps of phase shift by a 4 steps of phase shift, which can meet the needs of speed and accuracy of dynamic 3D reconstruction.