摘要
提出一种基于主成分分析(PCA)与相似递归残差补偿的人脸超分辨率算法。基于PCA获得高低分辨率人脸图像特征空间的映射系数,通过该系数重建初步的高分辨率人脸图像。利用高低分辨率人脸图像空间同一区域图像块的内容相似性,递归计算残差补偿图像。采用该残差图像对初步重建的全局人脸进行细节补偿。实验结果表明,该算法的重建效果较优。
This paper proposes a face super-resolution algorithm based on Principle Component Analysis(PCA) and similar recursive residue compensation. The mapping coefficients of low-resolution facial space are obtained based on PCA and preliminary face is reconstructed through these coefficients. Using the similar contents of the same face region in high and low resolution face image, a residual image is computed by recursive linear combination. It uses the residue image to compensate the global image reconstructed. Experimental results show that the proposed method produces high quality images.
出处
《计算机工程》
CAS
CSCD
2012年第13期196-198,共3页
Computer Engineering
基金
国家自然科学基金资助项目(61101215)
中央高校基本科研业务费专项基金资助项目(CHD2011JC146)
长安大学基础研究支持计划专项基金资助项目
关键词
人脸图像
超分辨率
递归
主成分分析
残差补偿
face image
super-resolution
recursion
Principle Component Analysis(PCA)
residue compensation