基于改进AP聚类算法的人脸标注技术研究  

Research on Annotation Technology of Face based on Improved AP Clustering Algorithm

作　　者：刘胜宇 刘家锋 黄庆成 唐降龙 LIU Shengyu, LIU Jiafeng, HUANG Qingcheng, TANG Xianglong （School of Computer Science ＆ Technology, Harbin Institute of Technology, Harbin 150001, China）

机构地区：哈尔滨工业大学计算机科学与技术学院,哈尔滨150001

出　　处：《智能计算机与应用》2011年第1X期35-38,共4页Intelligent Computer and Applications

基　　金：基金项目国家自然科学基金资助项目（61073128）,哈尔滨工业大学科研创新基金资助项目（HIT.NSRIF.2009158）.

摘　　要：利用网络新闻图像以及与图像相关的文字说明构造一个有人名标注的人脸库。人脸库中的人脸通过对新闻图像进行人脸检测得到，人名从与图像相关的文字说明中提取得到。通过一个聚类过程得到人脸与人名之间的对应关系，并通过一个合并聚类的过程可以识别出同一个人的不同名字。对AP聚类算法进行改进，实验表明相比于k-means与AP算法，改进的AP聚类算法能充分利用网络新闻图文互补的特性，从而得到更准确的人脸与人名对应关系。In this paper, a face dataset labeled by names is built from network news photographs and their associated captions. Faces are extracted from the images and names are extracted from the associated captions. A clustering process can find the correspondence between faces and associated names. A cluster merging process can identify the same person with different names. AP clustering algorithm is improved and experiments show that compared with k-means clustering algorithm and AP clustering algorithm, the improved AP clustering algorithm can make full use of the complemen- tary feature of images and captions in the network news to get more accurate correspondence between the faces and names.

关 键 词：网络新闻 人名 人脸图像 人脸标注 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术;自动化与计算机技术—计算机科学与技术]