The polarimetric signature of the sea surface is calculated, by using the small perturbation method, from a two-scale sea surface model. This signature is formalized by the Mueller matrix, whose coefficients are functions of the speed of the wind, the incidence of illumination and the azimuthal direction of the wave with respect to the direction of the wind. The polarimetric signature of the ocean is then described by the nine Huynen coefficients. In this paper, the cross-correlation between these coefficients is shown; a reduction of the polarimetric information to a Huynen vector whose dimension is minimal (five) is proposed. From this defined vector a Bayesian estimator (a maximum likelihood ratio test) is built, making it possible to estimate the wind direction and speed from polarimetric measurements of the sea backscattering.