摘要
提出了一种新的且有效的单幅图像超分辨率算法,该算法是基于自然图像中的局部缩放不变性原理。算法不需要额外的学习数据库,相反,只是从输入图像中的一个局部区域学习低分辨率图像块与高分辨率图像块之间的映射关系,因此很大程度上缩短了学习时间。为了能够得到更真实自然的图像结果,算法利用纹理上下文来搜索得到高分辨率图像块。实验证明,算法可以得到比较理想的效果。
A new effective single-image super-resolution method is proposed in this paper,which is based on the local scale invariance in natural ima- ges. It does not rely on an external example database. Instead ,it learns the mapping between low-resolution and high-resolution image patches from extremely localized regions in the input image, so it reduces considerably the learning time. To produce photo-realistic results, it exploits texture context to search for the high-resolution patch. Experimental results show its effectiveness.
出处
《电视技术》
北大核心
2014年第13期24-27,共4页
Video Engineering
关键词
纹理约束
局部自学习
单幅图像
超分辨率
局部缩放不变性
纹理上下文
texture-constrained
local self-examples
single image
super-resolution
local scale invariance
texture context
