基于相关分析和显著性检测的图像缩放方法  

Image Scaling Based on Correlation Analysis and Saliency Detection

作　　者：孙雯雯[1] 陶胜[1] SUN Wen-wen;TAO Sheng(Department 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：《宜春学院学报》2020年第12期53-57,共5页Journal of Yichun University

基　　金：福建省教育厅(项目编号:JT180262)。

摘　　要：对于非等比例缩放图像的问题,提出一种基于相关分析和显著性检测的图像缩放方法。首先将数字图像的列/行看作随机变量,利用相关分析的理论得到各列/行之间的相关系数,再与显著性检测相结合,定义各列/行的影响度值,标注影响度值低的列/行,使缩放只发生在影响度值低的列/行上。另外,采用影响度扩散的方法,将被缩放列/行的影响度转移到相邻两列/行,以避免因连续过多地缩放列/行而导致的整体概貌失真。实验表明:该缩放方法可以较好地保护图像重要内容,并能保持图像的整体结构,有较好的视觉效果。To solve the problem of non-uniform scale image,an image scaling method based on correlation analysis and saliency detection is proposed.Firstly,the column/row of digital image is regarded as random variable,and the correlation coefficient between each column/row is obtained by using the theory of correlation analysis.Then,the influence value of each column/row is defined and the column/row with low influence value is marked so that the scaling only occurs on the column/row with low influence value.In addition,the influence degree diffusion method is used to transfer the influence value of the scaled column/row to the adjacent two columns/rows.Experimental results show that the method can protect the salient content of the image and has good visual effect.

关 键 词：相关分析 显著性检测 图像缩放 影响度值 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]