面向人物图像的近似检索算法性能评测  

Performance Evaluation of Near-duplicate Web Image Retrieval Algorithm of Person Images

作　　者：罗玉梅[1] 王瑞 LUO Yumei;WANG Rui(Faculty of Compute Information,Kunming Metallurgy College,Kunming 650033,China)

机构地区：[1]昆明冶金高等专科学校计算机信息学院,云南昆明650033

出　　处：《昆明冶金高等专科学校学报》2018年第3期59-65,共7页Journal of Kunming Metallurgy College

摘　　要：最新研究表明,互联网上存在的海量近似重复图像可以为解决一些传统上很困难的计算机视觉任务提供新的解决方案。介绍了4个主流的用来进行近似重复图像检索的方法:Hash码,Mean SSIM,SIFT视觉词袋模型(Bo VW)和属性关系图(ARG)。构建了一个包含24 762幅图像的与人物有关的图像数据集,观察数据确定了4种近似重复图像类型。利用该数据集通过实验定量评估了4种近似重复图像检索方法的运行效率和检索精度,最后推荐使用基于SIFT特征的视觉词袋方法来进行面向人物图像的近似重复Web图像检索。Recent studies have revealed that numerous near-duplicate Web images can be used as an intermediate step to implement some traditional difficult computer vision tasks. This paper presents a comprehensive study of the existing near-duplicate image retrieval methods in a structural way. Four representatives of the existing methods,i. e. hash signature,mean SSIM,Bo VW with SIFT features and ARG,are experimentally evaluated using a self-constructed dataset containing 24 762 images. The experimental results reveal that compared with global feature based methods,local feature based ones are usually more appropriate for the task of person identification in web images,as they can deal with partial duplicate and scene similar images better. In particular,Bo VW with SIFT features is recommended as it provides the best trade-off between on-line speed and retrieval accuracy.

关 键 词：检索 WEB图像 算法 评测 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]