基于PCA的内容保持图像缩放方法  

Content-aware Image Scaling Based on PCA

作　　者：孙雯雯[1] 陶胜[1] 胡明颖[1] SUN Wenwen;TAO Sheng;HU Mingying(College of Science,Jimei University,Xiamen 361021,China)

机构地区：[1]集美大学理学院,福建厦门361021

出　　处：《新乡学院学报》2020年第12期28-32,共5页Journal of Xinxiang University

基　　金：福建省教育厅资助项目(JT180262)。

摘　　要：针对图像的非等比例缩放问题,提出了一种基于主成分分析的内容保持图像缩放方法。将数字图像的列/行看作是图像数据的特征维度,利用PCA算法确定其主成分,定义并计算各列/行的影响度,标注影响度低的列/行,让图像缩放发生在影响度低的列/行上。在标注列/行时采用了影响度扩散的方法,从影响度最低的列/行开始,每标注一列或一行,就将该列/行的影响度扩散到邻近列/行,避免了因连续缩放列/行像素而导致的整体概貌失真。试验结果表明:该缩放方法既能保持图像主体内容,也能兼顾图像整体概貌。Aiming at the problem of non-uniform image resizing,a content-aware image scaling based on principal component analysis(PCA)is proposed.Taking the column/row of digitalimage as the feature dimension of image data,the PCA algorithm is used to determine its principal component.Define and calculate the influence weight of each column/row,label the column/row with low influence weight,so that image scaling occurs on the column/row with low influence weight.When labeling the column/row,the influence weight diffusion method is adopted.Starting from the column/row with the lowest influence weight,the influence weight of the column/row is transferred to the adjacent column/row to avoid the overall profile distortion caused by the continuous scaling of column/row pixels.The experimental results show that the method can not only keep the main content of the image,but also take into account the overall picture.

关 键 词：主成分分析 图像缩放 内容保持 影响度 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]