Multi-band SAR images fusion using the EM algorithm in contourlet domain

Aimed at the fusion of multi-band Synthetic Aperture Radar(SAR) images, a new fused method using estimation theory in the contourlet transform domain is presented. Contourlet transform is a new "true" two-dimension presentation for images which provided a flexible multiresolution, anisotropy and directional expansion. The coefficients of contourlets can be accurately modeled by Gaussian mixture model. This approach is based on an image formation model which the contourlet coefficients of multi-band SAR images are described as the true scene corrupted by Gaussian mixture distortion. A set of iterative equations are developed using the Expectation Maximization(EM) algorithm to estimate the model parameters and produce the fused images. The efficiency of this approach is verified by fusing the Ku, X bands and L, Ku bands SAR images. Also, some statistical factors are employed for evaluating the objective quality of the fused result.