The method implemented in class VariationalRefinementImpl (file modules/video/src/variational_refinement.cpp) is a modification from [T. Brox, A. Bruhn, N. Papenberg, J. Weickert. High Accuracy Optical Flow Estimation Based on a Theory for Warping. ECCV (4) 2004: 25-36]
The OpenCV implementation has additional parameters, that do not appear in the original formulation of Brox et al : \delta and \zeta
The OpenCV implementation minimizes : \delta \Psi(E_I) + \gamma \Psi(E_G) + \alpha \Psi(E_S) where E_I,E_G,E_S are color constancy, gradient constancy and smoothness terms respectively. So, the purpose of this additional \delta is clear for me. However, I do not understand the purpose of the additional \zeta, which is used in the computation of the data term. Any comments on this?
Thanks for your time