Calibration of a pseudo-overlapped imaging based multiview system for high-resolution dynamic test

Abstract

An effective calibration method for a pseudo-overlapped imaging system was investigated based on pseudo-overlapped imaging. To address the restriction between field-of-view (FOV) enlargement and spatial resolution enhancement in optical monitoring, a multiview digital image correlation system based on pseudo-overlapped imaging was developed to reconstruct large movements of relatively small objects. Using a specially designed chessboard pattern, the intrinsic and extrinsic parameters of the four virtual cameras were calibrated without the need for overlapped field, which can be used in discontinuous FOVs. Consequently, the world coordinate systems of the different FOVs were reconstructed and unified. The results show that the calibration accuracy, with different stitching areas, had an error of 0.05 pixels, comparable to that of traditional multicamera systems. The effectiveness of the calibration method and the feasibility of the multiview measurement system based on overlapped imaging were validated through cylindrical translation experiments and high-speed dynamic experiments involving a ball.