This paper investigates the fusion of several kinds of SAR (Synthetic Aperture Radar) images with optical images, which were used as inputs for land cover classification. Once the images are from different sources, it must be corrected and referenced one over the other. The technique applied to perform the fusion was the classical IHS (Intensity-Hue-Saturation), where the I component was replaced by a product of polarimetric SAR images, that include: HH and HV polarization amplitude SAR image, the ratio of HV to HH amplitudes, the bands ratio and the Freeman-Durden Decomposition components. After fusing, all real images were classified by region growing method using the Bhattacharya distance. The complex image, i. e., SAR images, were classified by the same method, but using the Wishart distribution and the Bhattacharya distance. The classification accuracy of each method was measured by the Kappa coefficient.