An Improved Calibration Method for the Structured Light System Based on Self-correction of Reprojection Error

Abstract

Structured light 3D reconstruction is widely applied in various fields. During the calibration process of structured light 3D reconstruction, it is very important to raise the calibration accuracy on the parameters of the structured light system. In this paper, we propose a method based on re-projection error self-correction to obtain more accurate corner positions by screening the re-projection error values of DMD images. Experimental results show that this method can improve the calibration accuracy by 64.17%. We also propose an effective standard for the placement of calibration plate, which is of great significance to reduce the number of iterations of the program. According to a series of experiments based on the above standard, the number of iterations of the proposed re-projection error self-correction method is no more than 5 times. It proves that the proposed self-correction method and placement standard are feasible, the calibration process of structured light 3D reconstruction is optimized, and its calibration efficiency is improved.