Geometric Calibration of Digital Cameras. Part Ⅱ: Internal and External Geometry From Multi-View Alignment

摘要

This paper presents an algorithm for estimating the internal and external geometry of a digital camera by aligning a set of different views of a calibration plate with a convenient virtual image. The task is formulated as a nonlinear optimization problem where an appropriate cost function has to be minimized; the views are first prealigned with the reference image by means of the algorithm of Part Ⅰ. This initialization together with the closed-form computation of the gradient of the cost function allows the Levenberg-Marquardt algorithm employed to solve the nonlinear least-squares problem to rapidly converge to the optimal solution. The performance of this calibration algorithm is tested with a set of real images and discussed in the paper. Also, its accuracy is assessed by means of synthetic versions of the real images generated with the estimated parameters.