一种自适应正则化参数和模数的图像去卷积方法

自适应正则化方法是解决图像去卷积问题中平滑噪声和保持边缘矛盾的一种方法，传统的自适应正则化方法只考虑到图像的整体信息而忽略了图像内部不同区域的细节信息，本文提出的自适应正则化方法，利用Katsaggelos自适应正则化参数，自适应的改变正则化参数，并通过图像内部不同区域的信息自适应改变正则化模数矩阵．从而实现了平滑噪声和保持边缘的平衡，既保持了边缘又抑制了噪声，取得了更好的去卷积效果．

One method of image deconvolution of adaptive regularization parameter and modulus

The adaptive regularization method is to solve the conflict problem between smoothing noise and the edge-preserving of image deconvolution. But the traditional adaptive regularization method only takes into account the overall information of image while ignoring the details information of different areas within the image. The paper presents an adaptive regularization method to modify regularization parameter self-adaptively by manipulating Katsaggelos adaptive regularization parameter, and to modi- fy regularization modulus self-adaptively based on the information from the different internal regions of the image. Consequently we can achieve a goal smoothing noise and maintaining edge. That is to say, our method can reach a balance of edge-preserving and noise suppression, and also achieve a better deconvolution result.