摘要
提出了一种有效的高分辨率图像复原方法,将单幅图像的超分辨率复原转换到小波域中,对小波域的3个高频信息块分别进行处理,再通过基于学习的超分辨率复原方法来实现单幅图像的复原。实验表明,通过该算法恢复的高分辨率图像具有更好的视觉效果与峰值信噪比。
The paper shows an effective high-resolution image recovery tion recovery to the wavelet domain, and processes the three high-frequency ing the examples-based super-resolution method to achieve the recovery shows that the high-resolution images recovered by this algorithm has better method, which changes the single image super-resolu- information of the wavelet domain respectively. And us- of the single image super-resolution. The experiment visual effects and peak signal-to-noise ratio.
出处
《微型机与应用》
2013年第18期40-43,46,共5页
Microcomputer & Its Applications
基金
国防技术基础研究项目(科工技【2011】869号)
国防预研基金项目(B3120110005)
四川省教育厅项目(11ZA130)
关键词
小波变换
基于学习
自相似性
超分辨率
wavelet transform
example- based
patch redundancy
super-resolution
