Accurate Camera Calibration with Simplified Patterns with a Self-Calibration Approach

In this paper we present a unified approach for calibration and self-calibration of multi-camera systems. The technique is based on an algorithm for the resolution of the self-calibration problem, with which it is possible to find the camera geometry up to an arbitrary perspective transformation of the calibrated scene space. The correct geometry is then obtained by properly exploiting the dimensional constraints introduced with the observation of a very simple calibration pattern. Experimental results have been carried out with many different 3D configurations of the fiducial points, in order to calculate both the minimum number of points, necessary to reach the desired accuracy, and their best disposition in the scene space. Moreover, the accuracy of the proposed calibration method is evaluated and experimentally validated.