基于图像局部信息聚类的增强RAISR超分辨率算法  

Enhanced RAISR Super Resolution Algorithm Based on Local Information Clustering

作　　者：孙孟川 谭生详 王晔 SUN Mengchuan;TAN Shengxiang;WANG Ye(College of Computer Science,Sichuan University,Chengdu 610065)

机构地区：[1]四川大学计算机学院,成都610065

出　　处：《现代计算机》2021年第9期63-66,共4页Modern Computer

基　　金：国家重大项目(No.GJXM92579)。

摘　　要：在信息水平高度发达的今天,图像是信息承载的主要方式之一,人们对于高质量图像的需求也日益增加,然而在手机、电视等设备中难以运行复杂度高的超分辨率算法,因此需要一个快速、精确的超分辩率算法来集成到低性能的图像显示设备中。提出的方法是在RAISR算法的基础框架上,引入图像局部信息聚类的方法,在不过多增加算法复杂度的前提下增强RAISR方法的超分辨率质量,并且在Set5和Set14两个数据集上得到验证。Information technology has been highly developed today.Image is one of the main ways of information carrying,and people's demand for high-quality images is also increasing.However,it is difficult to run high-resolution algorithms in mobile phones,television and other de⁃vices.Therefore,a fast and accurate super-resolution algorithm is needed to integrate into low-performance image display devices.Based on the RAISR algorithm,the local information clustering method is introduced to enhance the super-resolution quality of the RAISR meth⁃od without increasing the complexity of the algorithm.Finally,it is verified on two datasets of set5 and set14.

关 键 词：超分辨率 RAISR 聚类 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]