超分辨率复原方法相关原理研究

介绍了超分辨率复原方法的概念和理论基础;重点总结了常用的超分辨率复原方法,并对相关的理论依据、优缺点和适用范围进行了详尽分析;对超分辨率复原方法的未来发展进行了展望.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,因而导致难以使用图像的先验信息进行超分辨率复原.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.

Correlation theory of super-resolution restoration method

Firstly, the basic concepts and theories of super-resolution restoration method are introduced. Secondly, some applications focused on common method of super-resolution restoration are summarized. Their theoretical basis, advantages and disadvantages, and scope of applications are exhaustively analyzed. Finally, the future development of super-resolution restoration method is prospected. 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 and doesn't have a general. Meanwhile it is difficult to use the priori information of the image to help super-resolution restoration. With space domain recovery method, a complex motion model can be easily established considering almost all of the imaging degradation factors, including noise, down sampling, fuzzy caused by non-zero aperture, degradation of optical system, and motion blur. As the same time, we could also add more perfect priori knowledge. Compared to the frequency domain method, space domain super-resolution restoration model is more close to actual degradation processes and is currently the most widely used super-resolution restoration method.