基于数据驱动紧框架的图像恢复  

Data-driven tight frame based image restoration

作　　者：杜淋 王聪[1] 

机构地区：[1]河海大学理学院,江苏南京211100

出　　处：《电子设计工程》2017年第22期178-181,185,共5页Electronic Design Engineering

基　　金：中央高校基本业务科研费专项基金(2015B38014;2015B19514)

摘　　要：提出了一种基于数据驱动紧框架的变分模型用于含有高斯脉冲混合噪声的图像恢复。该模型由包含L1-L2范数的拟合项和包含L1范数的光滑项构成。然后,又提出了ALM-APG算法用于解该模型。最后,数值实验表明所提出的模型与算法能够有效地进行图像恢复并且具有一定的实用性。This paper proposed a data-driven tight based variational model to deal with image restoration with mixed Gaussian-impulse noise. This model consists of the fitting term with L1-L2 norm and the smoothing term with L1 norm. Then,we also proposed ALM-APG algorithm to solve this model. Finally,supporting numerical experiments were preformed,which verified that our model and algorithm are able to remove mixed Gaussian-impulse noise.

关 键 词：图像恢复 脉冲高斯混合噪声 数据驱动紧框架 ALM-APG算法 

分 类 号：TN919.8[电子电信—通信与信息系统]