An optimized two-step camera calibration method
- Resource Type
- Authors
- H. Bacakoglu; Mohamed S. Kamel
- Source
- ICRA
- Subject
- Matrix (mathematics)
Mathematical optimization
Transformation (function)
Quaternion algebra
Calibration (statistics)
Rotation matrix
Quaternion
Algorithm
Mathematics
Camera resectioning
Nonlinear programming
- Language
An optimized two-step camera calibration algorithm is developed. The proposed method starts with the well known linear calibration which approximates the transformation as a 3/spl times/4 matrix. Based on the results of the linear calibration and the camera model we construct the 4/spl times/4 homogeneous transformation matrix. Quaternion algebra is used to extract the optimum rotation matrix and this optimization is later extended to the other calibration parameters. Our calibration method includes nonlinear optimization which takes into consideration lens distortions. The convergence rate of the nonlinear optimization is accelerated by three more objective functions we introduced. To assess the accuracy of our proposed quaternion method, Euclidean norm of the error matrix between the original and computed homogeneous transformation matrices is calculated and compared to those of the existing methods. Simulations show that the quaternion method yields more accurate results both before and after the nonlinear optimization.