基于显著图与主成分分析的图像缩放法研究  

Image Scaling Based on Principal Component Analysis and Saliency Detection

作　　者：胡明颖[1] 陶胜[1] HU Mingying;TAO Sheng(Jimei University,Xiamen 361021,China)

机构地区：[1]集美大学理学院,福建厦门361021

出　　处：《洛阳理工学院学报（自然科学版）》2021年第1期64-70,共7页Journal of Luoyang Institute of Science and Technology：Natural Science Edition

基　　金：福建省教育厅基金项目(JT180263).

摘　　要：为保证图像缩放过程中重要区域的完好性和整体视觉的连贯性,提出一种基于显著图与主成分分析的图像缩放方法。对图像各行各列进行主成分分析,提取最能表征图像重要内容的特征区域,并结合显著图的显著性信息,定义行、列向量的重要度,对非重要区域的行、列缩放,并进一步优化连续缩放行、列的选取。实验结果表明:基于显著图与主成分分析的图像缩放方法运行高效,效果良好。In order to meet the requirements of visual quality in image scaling,a method based on saliency image and principal component analysis is proposed.Initially,the saliency region which can best represent the content of the image is extracted by the principal component analysis of each row and column of the image.Simultaneously,the rows and columns of non-important regions are scaled.Followed,the selection of continuously scaled rows and columns is further optimized.The experimental results show that the combination of saliency image and principal component analysis is an efficient and content-based unequal image scaling method.

关 键 词：图像缩放 主成分分析 显著图 重要度 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]