Abstract: To overcome the big error existing in the common optical flow technology for determining wind direction based on brightness constancy model (BCM), an energy function optimization method was proposed for determining wind direction which depends on brightness and gradient constancy model (BGCM) with the feature of sea clutter images. Firstly, energy function is constructed with the common brightness constancy assumption, gradient constancy assumption as well as spatio-temporal smoothness constraint. Secondly, by calculating the derivation of optical flow components from energy function, the nonlinear Euler-Lagrange equations are obtained. Finally, optical flow field and main ocean surface wind direction can be achieved with iterative method to solve Euler-Lagrange equations. The improved method is finally validated with the real shore based X-band radar image. The superior performance of the proposed method is demonstrated by smaller estimation errors and well correlation when compared with results of the previous technique.

Key words: optical flow method, energy function, ocean surface wind direction, X-band radar, sea clutter image