Satellite image blind restoration based on surface fitting and iterative Multishrinkage method in redundant wavelet domain

In view of the non-unique, unstable and even divergent properties of the solution of iterative blind restoration, a non-iterative blind image-restoration algorithm was proposed, which included MTF estimation step and image-restoration step. In the MTF estimation stage, based on every degradation process of satellite imaging chain, a combined parametric model of MTF was given and used to fit the surface of normalized logarithmic amplitude spectrum of degraded image. In the image-restoration stage, a multivariate statistical model was introduced that can perfectly characterize the dependency of the neighboring wavelet coefficients and of reference coefficient and its parent. During the image-restoration, optimization-transfer method was adopted to decompose the image-restoration procedure into two simple steps: Landweber iteration and Multishrinkage denoising. In the numerical experiment, the comparison of restored panchromatic SPOT-5 (5m) images with corresponding panchromatic SPOT-5 (2.5m) image and higher resolution aerial image from the Google Earth software shows that the proposed algorithm can significantly restore some detailed information and effectively inhibit noise.