提出一种新的利用参考图像的模糊估计与恢复方法。同一场景不同时相的图像, 由于拍摄条件的不同, 清晰化程度也往往不同, 将高质量的图像作为参考, 为模糊估计和恢复提供先验知识, 在Bayes统一的框架内进行快速PSF估计, 然后使用总变分最小化方法进行恢复。通过真实遥感图像实验, 结果表明所提出的方法可以有效地估计退化模糊、恢复未失真图像。

Remote sensing images are sometimes corrupted by blur and noise. To solve this problem, a novel blur estimation and restoration approach is presented in this paper. The approach uses a high quality image of the same scene as a reference. Such a reference image is usually available, given the increasing popularity of remote sensing applications. With the reference image, the point spread function (PSF) of another more blurry image can be less difficultly and more accurately estimated in the Bayesian framework. Once the PSF is known, many deconvolution approaches can be employed such as the total variation minimization method, which was used in this paper. Experiments with real remote sensing images show that the proposed method can effectively estimate the PSF and restore the blurred image.