文献类型：期刊文献

中文题名：老龄面部数据抽取的肤色显著性方法

英文题名：Aged facial data extraction method by using skin color saliency

作者：戴国庆[1];张晟磊[1];袁玉波[1,2]

机构：[1]华东理工大学信息科学与工程学院,上海200237;[2]上海大数据与互联网受众工程技术中心,上海200072

年份：2022

卷号：42

期号：S02

起止页码：217

中文期刊名：计算机应用

外文期刊名：journal of Computer Applications

收录：CSTPCD;;北大核心:【北大核心2020】；CSCD:【CSCD_E2021_2022】；

语种：中文

中文关键词：肤色显著性;遗传算法;老龄面部;人脸特征;人脸轮廓

外文关键词：skin color saliency;Genetic Algorithm(GA);aged face;facial feature;face contour

摘要：随着老龄化日益严重,相应数据驱动的决策支持系统成为智慧养老的关键,而人脸数据是老龄大数据的重要组成部分,如何精准抽取老龄面部数据是一个重要的技术难题。基于此,提出肤色显著性的概念,并将其应用到老龄面部数据抽取。首先,建立不同人种、不同年龄阶段的皮肤参考数据库,得到不同人种的老龄的肤色参数,建立肤色显著性模型;同时,利用肤色参数抽取脸部肤色前景,并利用肤色显著性抽取的种子区域;最后,采用81人脸特征点和形态学特征,确定人脸轮廓,并利用遗传算法(GA)抽取得到老龄面部数据。实验结果说明,所提方法能显著提升抽取准确率,在SPFace以及UTKFace的老龄数据上的准确率达到80%以上,具有很好的应用前景。
As aging becomes more and more serious,the corresponding data-driven decision support system has become the key to smart elderly care.Facial data is an important part of aged big data,and how to extract aged facial data accurately is an important technical problem.Based on this,the concept of skin color saliency was proposed and applied to the extraction of aged facial data.First,a skin reference database of different races and different ages was established to obtain the skin color parameters of different races and a skin color saliency model was constructed.At the same time,the skin color parameters were used to extract the foreground of facial skin color,and the skin color saliency was used to extract the seed region.Finally,eighty-one facial feature points and morphological features were used to determine the contour of the face,and the Genetic Algorithm(GA)was used to extract the aged facial data.Experimental results show that the proposed method significantly improves the accuracy of extraction,and the accuracies on SPFace and UTKFace aged data reach more than 80%,which has good application prospect.