Multicamera Systems: Calibration and Applications

Multi-camera calibration is a crucial step for 3D reconstruction applications. When the accuracy of the reconstruction is the main goal, the size and the measurement's accuracy of the target-set used as a test scene for calibrating the cameras can pose serious limits to the flexibility and the cost of the calibration strategy. Furthermore, parameter drifts due to mechanical shocks or thermal changes may impair the validity of the estimated camera parameters during the acquisition session. In this article we present a low-cost, accurate and flexible approach to the estimation of multicamera system parameters. We show how to drastically simplify the calibration target-set and avoid costly high-accuracy measurements on it, without giving out accuracy in the results. Furthermore, we show how to make the calibration process adaptive through a process of on-the-fly parameter correction based on the analysis of natural scene features.