Research on global measurement method based on multi-cameras

Abstract

The 3D reconstruction based on multi-cameras has been widely used and studied in the field of advanced manufacturing. Many 3D reconstruction methods have been proposed, but the limitation of high-accuracy, large-scale measurement always exists. The construction and calibration of multi-cameras system are the fundamental task as well as the key for the accurate measurement. In this paper, a high-precision camera calibration method for existing multi-cameras system is put forward. The transparent glass checkerboard and the principle of least time is applied in the method. The application of the transparent glass checkerboard leads to refraction phenomenon, which will generate calibration error. The refractive projection model is proposed to eliminate the error. Then the Scale-invariant feature transform (SIFT) algorithm is used to extract the feature point and epipolar geometry constraint is used to match the feature points. Finally, the 3D points can be obtained based on the principle of the triangle. The experimental results show the ability and robustness of the proposed approach.