图像超分辨率复原方法及应用  

Methods and Applications of Image Super-Resolution Restoration

作　　者：陈健[1,2,3] 高慧斌[1] 王伟国[1] 毕寻 

机构地区：[1]中国科学院长春光学精密机械与物理研究所,吉林长春130033 [2]中国科学院大学,北京100049 [3]吉林大学通信工程学院,吉林长春130012

出　　处：《激光与光电子学进展》2015年第2期41-50,共10页Laser & Optoelectronics Progress

基　　金：吉林省重大科技公关专项(20126015)

摘　　要：介绍了超分辨率复原方法的概念和理论基础,重点总结了一些常用的超分辨率复原理论的国内外研究现状,并对它们的理论依据、优缺点和适用范围进行了详尽分析,对超分辨率复原理论的应用领域进行了介绍。超分辨率复原方法分为频域法和空域法。频域复原法原理简单清楚,计算方便,但是所建立的运动模型都是平移模型,不具有一般性,同时难以利用正则化约束,这就导致难以使用图像的先验信息来帮助进行超分辨率复原。空域复原法可以很方便地建立复杂的运动模型,同时考虑了几乎所有的图像降质因素,例如噪声、降采样、由非零孔径时间造成的模糊、光学系统降质和运动模糊等,还可以加入更完善的先验知识,相比于频域复原法,空域超分辨率复原模型更符合实际的图像退化过程,是目前应用最广泛的一类超分辨率复原方法。The basic concepts and theories of super-resolution restoration method is introduced. Some applications focused on common method of super-resolution restoration is summarized. Their theoretical basis, advantages and disadvantages, and scope of applications are exhaustively analyzed. The applications of super-resolution restoration theory is introduced. Overall, the super-resolution restoration methods are divided into frequency domain method and space domain method. Frequency domain recovery method is simple in principle and easy in calculation. But its motion model is shift model have no generality. Meanwhile it is difficult to use the priori information of the image to help super-resolution restoration. Space domain recovery method can be easily taken degradation and motion blur. More perfect priori knowledge is added. Compared with the frequency domain method, space domain superresolution restoration model is more close to actual degradation processes and currently the most widely used superresolution restoration method.

关 键 词：图像处理 超分辨率复原 频域法 空域法 凸集投影约束 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]