Abstract: Considering the big error exits in traditional gray model optical flow method for calculating ocean surface wind,a novel method is proposed,which consists of two parts.Firstly,Ocean wind direction retrieval algorithm based on brightness and gradient constancy model according to the characteristic of sea clutter was proposed.Furthermore,main wind direction can be achieved by analyzing optical flow field.The other is wind speed retrieval algorithm.By applicatingBP neural network for constructing model among wind speed,radar cross section,radio direction and wind direction,wind speed can be derived.Finally,simulated and real radar images are used for experiments,and then we compare the measured wind vector with real wind vector.The results show that the X-band radar does a good job in measuring wind vector,and the wind vector reversion algorithm has a good feasibility.