摘要
单幅彩色图像进行超分辨率重建,一般先对亮度分量Y进行超分辨率重建,再对色度分量U和V进行简单插值,重建图像色彩模糊。针对此问题,文中提出一种同时对亮度与色度分量进行基于邻域嵌入的彩色图像超分辨率重建算法,该算法有效利用了色度分量的先验信息。为提高算法效率,使用K均值聚类的方法对样本集进行分类,并使用二叉树搜索方法确定样本类别。实验结果表明,文中提出的算法不仅提高了彩色图像的重建质量,并有效降低了算法的运行时间。
For single-frame color image super-resolution reconstruction,most techniques use super-resolution reconstruction only on Y-channel. Directly use interpolation algorithms for the chrominance channels( U , V ) which decide the color. The color of the reconstructing image is fuzzy. In order to solve this problem,propose a new super resolution method of color image based on the neighbor embed-ding. It uses the super resolution method to jointly estimate the luminance information and chrominance information,effectively using the chrominance components of a priori information. To enhance the efficiency of algorithm,use the K-Means clustering method to classify the sample set and apply the binary tree search method to determine the classification of sub-sample set. Experimental results show that the proposed method can effectively reduce the running time as well as improve reconstruction equality of low resolution color images.
出处
《计算机技术与发展》
2015年第6期25-28,共4页
Computer Technology and Development
基金
江苏省自然科学基金青年基金(BK20130867)
江苏省高校自然科学研究项目(13KJA510004
12KJB510019)
南京邮电大学科研基金(NY212015
NY213133)
广州市软件和信息服务业发展专项资金(2060404)
关键词
彩色图像超分重建
邻域嵌入
色度分量
样本集分类
color image super-resolution reconstruction
neighborhood embedding
YUV color space
classification of sample-sets