带微调参量的正则化方法及其在降质图像恢复问题中的应用

A Regularization Method with Spinning Function Parameters and its Application in Restoring Degraded Image

1引言 降质图像恢复问题就是图像处理领域里一类反问题.降质图像恢复中的解通常是病态的,利用正则化方法恢复图像取得了较好效果[1～3].然而,传统的正则化方法中正则逆算子只含正则化参数,它不能充分地融合其它信息,得到的正则解逼近真解的效果不很理想.为了取得好的恢复效果,各种各样融合其它信息的方法提出来了,以使正则化方法的恢复效果更好.

Introduce spinning function parameters in regularized solution definition consequently regularization operators can keep more message from outside so that we can choose regularization operator which approaches true solution as well as possible. Based on the concept of spinning function parameters we discuss corresponding regularization method. By the instruction of the method in restoring degraded image ,we work out a model of implicit form of regularization operators which can carry some more message. When the spinning function parameters are constant functions there is quick algorithm to find the regularized solution and we give an explicit form of the regularized solution. The regularization method with spinning function parameters will have theoretical signification in solving other inverse problems.