基于模糊非线性回归算法的图像复原

为了利用多幅具有相似降质特性的退化图像信息恢复出原图像,提出了一种新的基于模糊参量非线性回归估计的图像盲复原算法。 该算法充分利用多幅图像具有的相似降质特征,首先给出一种非理想光照分布参数的模糊非线性回归估计方法,然后通过估计出的参数得到与原图同样大小的非理想光照分布图,最后通过观测图像与非理想光照分布图相消的方法,对退化图像进行复原。 实验结果表明,该算法运算快速,对边缘细节及平滑区域均有良好的修复能力,在修复效果上明显优于一般图像修复算法。

Image Restoration Based on Fuzzy Nonlinear Regression Algorithm

In order to restore image by using the multiframe degraded images of the similar degraded characters, a novel fuzzy parameter nonlinear regression algorithm for blind restoration was presented. The advantage of the similar degraded characters of multiframe images was fully taken. Firstly, the fuzzy value of illumination parameters was estimated through fuzzy nonlinear regression method. Then the nonideal distribution of illumination image was estimated by the parameter gained before. Finally, the degraded image was restored by the contraction between the observed image and the nonideal distribution of illumination image. Results of experiments demonstrated that this method works efficiently; it has better ability in restoring both smooth area and edge-contained area. Compared with others, this method in the repair effect is superior to the general image restoration algorithm.