On global parametric motion estimation with lens distortion correction

In this paper, two-dimensional (2-D) motion estimation techniques based on the well-known optical-flow equation are treated. Cost functions are defined for various motion models that allow the optical flow equation (OFE) to be used in an optimization framework for the purpose of estimating the 2-D motion parameters. A method of extending these cost functions to include camera parameters is then proposed. The resulting sometimes non-linear optimization framework takes into account motion-model parameter estimation in the presence of geometric lens distortions. Linearization, when necessary, results in an OFE-like motion estimation method.