摘要
图像超分辨率是指利用一幅或多幅低分辨率图像,运用相应的算法来获得一幅清晰的高分辨率图像.然而,传统的基于插值和重建的方法已很难获得进一步的突破.近年来出现的基于学习的方法为超分辨率的发展重新注入了活力.通过回顾插值、重建和学习这3个层面的超分辨率算法,分析了超分辨率技术的以往研究和最新进展,着重讨论了各算法在还原质量、通用能力等方面所存在的问题,并对未来超分辨率技术的发展作了一些展望.
The goal of super resolution is to get high resolution images from at least one low resolution image. However,with the traditional (interpolation-based and reconstruction-based) methods it is difficult to make further important progress. The emerging learning-based super-resolution methods breathe new life into the research. By reviewing these three kinds of methods,the history of super resolution was surveyed,the limitation of existing methods was discussed and the roadmap was analyzed for future developme...
出处
《山东大学学报(工学版)》
CAS
北大核心
2009年第1期27-32,共6页
Journal of Shandong University(Engineering Science)
基金
国家自然科学基金资助项目(60635030)
国家“863”高技术研究发展计划资助项目(2007AA01Z176)
关键词
图像处理
超分辨率
邻域嵌入
图像重建
image processing
super resolution
neighbor embedding
image reconstruction