Least-squares phase estimation with wrapped measurements and branch points.

A nonorthogonal model for 2D signals with rotational components is presented, which enables estimation of phase values from observations of its local gradients. In this research, the rotational components are caused by the presence of branch points, which indicates phase wrapping. Using the proposed model, the phase is estimated using standard least-squares or recently proposed wavelet techniques by processing a linear combination of the wrapped observed gradients and the curl generated by phase wrapping.