Implicit Camera Calibration Based on a Nonlinear Modeling Function of an Artificial Neural Network
ISNN '09 Proceedings of the 6th International Symposium on Neural Networks on Advances in Neural Networks
Accurate camera calibration using the collinearity constraint
CIRA'09 Proceedings of the 8th IEEE international conference on Computational intelligence in robotics and automation
Automatic on-the-fly extrinsic camera calibration of onboard vehicular cameras
Expert Systems with Applications: An International Journal
| Hi-index | 0.15 |
Accurate measurement of the position of features in an image is subject to a fundamental compromise: The features must be both small, to limit the effect of nonlinear distortions, and large, to limit the effect of noise and discretization. This constrains both the accuracy and the robustness of image measurements, which play an important role in geometric camera calibration as well as in all subsequent measurements based on that calibration. In this paper, we present a new geometric camera calibration technique that exploits the complete camera model during the localization of control markers, thereby abolishing the marker size compromise. Large markers allow a dense pattern to be used instead of a simple disc, resulting in a significant increase in accuracy and robustness. When highly planar markers are used, geometric camera calibration based on synthetic images leads to true errors of 0.002 pixels, even in the presence of artifacts such as noise, illumination gradients, compression, blurring, and limited dynamic range. The camera parameters are also accurately recovered, even for complex camera models.