文章基本信息

标题：Robust super-resolution using kernel regression with outliers-reduction scheme
本地全文：下载
作者：Feng Xu ; Lizhong Xu ; Fengchen Huang 等
期刊名称：Scientific Research and Essays
印刷版ISSN：1992-2248
出版年度：2011
卷号：6
期号：18
页码：3834-3844
DOI：10.5897/SRE11.436
语种：English
出版社：Academic Journals
摘要：In the process of recording a digital image, super-resolution (SR) is a feasible soft method for solving the limitation of device and effect of environment. During the last two decades, many researchers proposed various SR algorithms for image reconstruction. Among these algorithms, kernel regression is a helpful tool which considers not only spatial distance between center pixel and neighbor pixel but also structural information. However, the problem of removing noise and outlier in kernel regression can be further studied and solved. In this paper, we proposed a new idea of so called trilateral kernel regression. Besides the above factors considered, the new kernel regression with trilateral idea considers additional factor: Confident correlation of pixels, so it can obtain more accurate result. Experiments are carried out to demonstrate the effectiveness of our method. The index of RMSE can be reduced by 2 or even 4 in some severe case.
关键词：Super-resolution; kernel regression; trilateral kernel; outliers reduction; confident index