Multi-camera parameter tracking

Abstract

The quality of 3-D reconstructions with multi-camera acquisition systems is strongly influenced by the accuracy of the camera calibration procedure. In fact, when acquiring a long sequence of views, mechanical shocks, vibrations and thermal gradients could cause a significant drift of the camera parameters. The authors propose a method for tracking the camera parameters and, whenever possible, correcting them accordingly. This technique does not need any a priori knowledge or test objects to be positioned in the scene, as it exploits natural scene features. The approach is based on accurate detection, matching and back-projection of luminance corners and spots in the scene space. Such features are then tracked over time to detect unexpected parameter changes or drifts, and to apply corrections to them. Experimental results on real sequences are reported in order to prove the effectiveness of the proposed technique. It is shown that changes in the calibration parameter are correctly detected and, when this happens, the camera system can be re-calibrated with an accuracy that increases with the number of tracked feature points.