计算机工程 ›› 2019, Vol. 45 ›› Issue (2): 191-194.doi: 10.19678/j.issn.1000-3428.0050113

• 人工智能及识别技术 • 上一篇    下一篇

基于加权字典对学习的人脸年龄估计方法

赵军,侯凯艳,杨林   

  1. 重庆邮电大学 计算智能重庆市重点实验室,重庆 400065
  • 收稿日期:2018-01-15 出版日期:2019-02-15 发布日期:2019-02-15
  • 作者简介:赵军(1971—),男,教授、博士,主研方向为人工智能、模式识别、数据挖掘;侯凯艳、杨林,硕士研究生。
  • 基金项目:

    国家自然科学基金(61379114)。

Age Estimation Method of Face Based on Weighted Dictionary Pair Learning

ZHAO Jun,HOU Kaiyan,YANG Lin   

  1. Chongqing Key Laboratory of Computational and Intelligence, Chongqing University of Posts and Telecommunications,Chongqing 400065,China
  • Received:2018-01-15 Online:2019-02-15 Published:2019-02-15

摘要:

针对现有人脸年龄估计方法多数将人脸各部分同等对待或忽视部分特征的问题,提出一种基于加权字典对学习(DPL)的人脸年龄估计方法。将人脸进行分块,使用局部二值模式算法对人脸的主要特征区域和次要特征区域分别进行特征提取,得到人脸的主要特征和次要特征,再利用这2种特征分别训练DPL模型,并赋予不同的权重,使用训练好的加权DPL模型对目标人脸图像进行年龄分类。在MORPH和FG-NET数据集上的实验结果表明,该方法具有较高的分类准确率。

关键词: 年龄估计, 主要特征, 次要特征, 加权, 字典对学习, 分类

Abstract:

Aiming at the problem that the existing face age estimation method treats all parts of the face equally or ignores some features,a face age estimation method based on weighted Dictionary Pair Learning (DPL) is proposed.The face is segmented,and the main feature region and the secondary feature region of the face are extracted separately by using the local binary pattern algorithm to obtain the main features and secondary features of the face.The retraining DPL model uses these two features separately and is given different weights.The trained weighted DPL model is used to classify the target face images.Experimental results on MORPH and FG-NET datasets show that the method has higher accuracy.

Key words: age estimation, primary feature, secondary feature, weighted, Dictionary Pair Learning (DPL), classification

中图分类号: