An ocean surface wind direction retrieval method from multi-polarization airborne synthetic aperture radar images is presented in order to retrieve high precision wind direction SAR image and improve the accuracy of ocean surface wind. The method relies on the ability of two-dimensional (2D) continuous wavelet technique with better time-frequency than Fourier transform and local gradient, which combines 2D continuous wavelet and FFT to exact the ocean surface wind direction from airborne SAR images. The proposed method is executed by using several kinds of mother wavelet functions for the C-band co-polarization and cross-polarization airborne sounding images, and the retrieved ocean surface wind direction is compared with NCEP data and buoy data. The verification results show that the wavelet based ocean surface wind direction retrieval algorithm is suited to retrieve wind direction from airborne SAR sounding data. The choice of mother wavelet function has a certain influence on the results, and 2D Mexican-Hat is the best mother wavelet for wind retrieval. Both co-polarization and cross-polarization airborne sounding images are suited to retrieve ocean wind direction.