华南理工大学学报(自然科学版) ›› 2015, Vol. 43 ›› Issue (5): 51-58.doi: 10.3969/j.issn.1000-565X.2015.05.009

• 电子、通信与自动控制 • 上一篇    下一篇

基于构图规则的图像美学优化

王伟凝 刘剑聪 徐向民† 姜怡孜 王励   

  1. 华南理工大学 电子与信息学院,广东 广州 510640
  • 收稿日期:2014-07-17 修回日期:2015-02-25 出版日期:2015-05-25 发布日期:2015-05-07
  • 通信作者: 徐向民(1972-),男,教授,博士生导师,主要从事人机交互、计算机视觉和 EDA 设计研究. E-mail:xmxu@scut.edu.cn
  • 作者简介:王伟凝(1975-),女,博士,副教授,主要从事图像处理与模式识别、图像情感分析、图像分类与检索、机器学习研 究. E-mail: wnwang@ scut. edu. cn
  • 基金资助:

    国家自然科学基金资助项目(61171142);广东省战略性新兴产业核心技术攻关项目(2011A010801005);广州市科技计划珠江新星项目(2012J2200010);华南理工大学中央高校基本科研业务费专项资金资助项目(2014ZZ0036)

Aesthetic Enhancement of Images Based on Photography Composition Guidelines

Wang Wei-ning Liu Jian-cong Xu Xiang-min Jiang Yi-zi Wang Li   

  1. School of Electronic and Information Engineering,South China University of Technology,Guangzhou 510640,Guangdong,China
  • Received:2014-07-17 Revised:2015-02-25 Online:2015-05-25 Published:2015-05-07
  • Contact: 徐向民(1972-),男,教授,博士生导师,主要从事人机交互、计算机视觉和 EDA 设计研究. E-mail:xmxu@scut.edu.cn
  • About author:王伟凝(1975-),女,博士,副教授,主要从事图像处理与模式识别、图像情感分析、图像分类与检索、机器学习研 究. E-mail: wnwang@ scut. edu. cn
  • Supported by:
    Supported by the National Natural Science Foundation of China(61171142)

摘要: 随着数码相机和智能手机的普及,数字图像的获取越来越便捷. 然而普通用户由于缺少专业摄影知识,拍摄的照片往往难以获得满意的效果. 为此,文中提出了一种综合考虑图像主体与背景构图规则的图像自动美学优化方法,根据三分法则和视觉平衡法则,采用主体提取、区域分隔线检测、基于样例的图像修复方法、细缝裁减法、主体重置等方法,通过图像主体位置、大小及背景区域划分的调整对图像进行美化. 实验结果表明,此方法能够在保护图像中的视觉不产生形变且保持图像信息不丢失的基础上,实现图像的构图美化.

关键词: 图像增强, 图像美学, 构图优化, 基于样例的图像修复, 细缝裁减算法

Abstract: With the prevalence of digital cameras and smart phones,people can acquire digital images more and more conveniently. However,normal users are not satisfied with their photos constantly because they are not fami-liar with the professional knowledge of photography. In order to solve this problem,an approach of image aesthetic enhancement,which takes into consideration both the subject and the background layout of an image,is presented.
According to the rules of trisection and visual balance,this approach retargets the subject and resizes the certain part of background in an image to enhance image aesthetic quality by means of subject extraction,region borderline detection,exemplar-based image inpainting,seam carving and subject retargeting. Experimental results show that the proposed approach can enhance the aesthetics of image composition without deforming the salient objects in an image and without losing any image information.

Key words: image enhancement, image aesthetics, composition optimization, exemplar-based image inpainting, seam-carving algorithm

中图分类号: