摘要
针对三维人脸识别中人脸数据部分缺失、遮挡以及损坏等情况,以及由训练样本缺乏引起的单训练样本问题,定义基于局部关键点的多三角形统计特征,该特征不仅能够在人脸数据部分可见的情况下保证鲁棒性,在人脸表情和姿态变化时也能准确描述人脸。针对单训练样本问题,提出一种两阶段加权协同表示方法。将提取的人脸局部特征作为先验知识,计算基于类的概率估计,并将该概率估计作为第二阶段分类中的局部约束,进而提高识别性能。实验结果表明,该方法可有效提高单样本部分人脸的识别率。
3D face recognition with the availability of only partial data and single training sample is a highly challenging task. In order to address challenge ,this paper defines the statistical feature of multiple triangles based on local key points ,which is robust to partial facial data,large facial expressions and pose variations. Aiming at the single sample problem,this paper proposes a two-phase weighted Collaborative Representation（CR） classification method. A class-based probability estimation is calculated based on the extracted local descriptors as prior knowledge, and this probability estimation is used to be local constraint in the second stage of classification to enhance the discriminating ability. Experimental results show that the proposed method can improve the recognition rate in the case of the partial facial data and single training sample.
出处
《计算机工程》
CAS
CSCD
北大核心
2016年第9期144-150,共7页
Computer Engineering
基金
国家自然科学基金资助项目(61403265
61471371)
四川省科技计划基金资助项目(2015SZ0226)
关键词
三维人脸识别
三维表示
稀疏表示
部分面部数据
单样本
3D face recognition
3D representation
Sparse Representation（SR）
partial facial data
single sample
