空域中基于正则化的自适应超分辨率图像复原

对于模糊图像的复原问题,从正则化技术克服问题病态性的思想出发,研究了一种有效的超分辨率图像复原方法.在Nguyen等人的正则图像复原框架的基础上,根据Roberts交叉梯度算子构造正则项,从自适应的角度生成正则化参数,并用共轭梯度法求解该模型的目标泛函极小值.计算机仿真结果表明,该方法可较好的再现图像的重要信息,复原图像的相对误差降低,同时,峰值信噪比和主观视觉效果方面都有明显的提高.

Adaptive super-resolution image restoration based on regularization in spatial domain

Aiming at the restoration of blurred image,an effective approach to super-resolution image restoration is proposed based on regularization technique for dealing with ill-posed problem.Under the frame of the regularized super-resolution restoration which is pro-posed by Nguyen.Regularization item is choosen which based on the Roberts cross gradient operator and determines the regularization parameter adaptively.The conjugate gradient method is used to solve the minimum of the constraint equation.Computer sim...