基于轻量化网络的图像超分辨率研究综述

An overview of image super-resolution research based on lightweight networks

近几年通过深度学习来进行图像超分辨率的研究越来越多,但更多的研究在于通过改变网络结构的深度和宽度来提升图像超分的质量,很少有研究采用轻巧而有效的网络来提高超分的效率而不影响其性能。因此,本文主要阐述几种轻量化网络,对其原理进行阐述,并对未来图像超分的发展趋势进行了展望。

In recent years,more and more research has been conducted on image super-resolution through deep learning,but more research lies in improving the quality of image super-score by changing the depth and width of the network structure,and few studies have adopted lightweight and effective networks to improve the efficiency of super-score without affecting its performance.Therefore,this paper mainly describes several lightweight networks,analyzes their principles and advantages and disadvantages,and analyzes the current status of image super resolution and looks forward to the future development trend of image super resolution.