Automatic age estimation via sparse representation
首发时间:2012-02-22
Abstract:Automatic age estimation from face has received increasing attention due to its wide range of application. A successful age estimator typically consists of two key modules: age-related feature extraction and age estimation by regression or classification. In this paper we propose a novel age estimator method based on sparse representation. In the feature exaction stage, the mid-level Spatial-Pyramid face representation based on Sparse codes of SIFT features (ScSPM) is used to characterize the age-related variance. For age estimation, linear sparse regression models are learned which can not only select the most discriminative features but also perform the age estimation. The hierarchical strategy, which first coarsely classifies the faces into age groups and then finely estimates the detailed age by the linear regression model of this group, is adopted to deal with the non-linearity attribute of aging to improve the performance of the age regression model. To our best knowledge, it is the first time to apply ScSPM and sparse linear regression to age estimation. The experimental results show that the proposed approach outperforms the state-of-the-art on the FG-NET database and achieves competitive performance on the MORPH database.
keywords: Pattern Recognition Age Estimation Sparse Representation Spatial Pyramid Matching elastic net
点击查看论文中文信息
基于稀疏表示的自动年龄估计
摘要:基于人脸的自动年龄估计研究因其广泛的应用前景而越来越被人们所关注。一个成功的年龄估计系统主要包括两个部分:年龄特征的提取和基于分类或回归的年龄估计。论文提出了一种基于稀疏表示的年龄估计新方法。该方法基于SIFT特征的稀疏编码,采用中层稀疏空间金字塔人脸表示作为年龄特征,然后采用线性稀疏回归模型来同时选择特征并完成年龄的估计。针对人脸老化的非线性性,该方法采用层次模型,即首先训练若干个分类器将人脸粗分类到不同的年龄组,然后再在该年龄组中训练对应的线性模型进行精确年龄的估计。在FG-NET人脸库和MORPH人脸库上的实验结果验证了本文提出的方法的有效性和优越性。
基金:
论文图表:
引用
No.****
同行评议
共计0人参与
勘误表
基于稀疏表示的自动年龄估计
评论
全部评论0/1000