Estimation of surface velocity from satellite imagery with large-scale displacement and longer temporal interval is one of the most challenging remote sensing problems in ocean dynamics. In this paper, a nonlinear inverse model using the displaced frame central difference (DFCD) equation has been created for estimating velocity in an image sequence. Iterative equations with Gauss-Newton and Levenberg-Marquardt algorithms are formulated for solving the nonlinear system of equations. A unified adaptive framework is developed based on the DFCD equations, velocity field modeling, a nonlinear least-squares model, and an algorithm of progressive relaxation of the overconstraint for seeking a flow field in which each vector is consistent with its neighbors. A numerical model is used as a benchmark to examine the accuracy of this new technique. Three sequences of NOAA Advanced Very High Resolution Radiometer (AVHRR) images taken in the New York Bight are also used to demonstrate the performance of the proposed technique. The estimated velocity fields are compared with those measured with the coastal ocean dynamics applications radar array. The experimental results indicate that the new estimator with the DFCD equation has major improvement over the linear inverse model for the simulation and real AVHRR data sets.