结构信息和特征信息结合的素描人脸识别
首发时间:2017-09-22
摘要:素描人脸识别(SFR)在法律刑侦方面的广泛应用,吸引着越来越多的学者投身于此项研究。根据人脸识别认知理论(FRCT),人类在人脸认知过程中主要依靠两种信息-结构信息和特征信息。因此,本文提出了一种基于结构信息和特征信息的素描人脸识别方法。首先,使用HOG特征对整幅人脸图像提取结构信息。然后,在局部人脸元素上使用用稀疏核原型表示对特征信息进行描述。将结构信息和特征信息分别用于素描照片匹配,在决策级上进行融合,并通过最近邻分类器进行最终识别。本文算法在不同数据库下的实验结果表明该方法优于其他算法。
关键词: 素描人脸识别 结构信息 特征信息 认知理论 稀疏核原型表示
For information in English, please click here
Structural Information and Feature Information-based Sketch Face Recognition
Abstract:Sketch face recognition (SFR) has been widely and successfully applied in law enforcement, which attracts a growing number of researchers. According to Face Recognition Cognitive Theory (FRCT), human utilizes two kinds of information - structural information and feature information for face cognitive. Thus, this paper proposes a SFR approach based on structural information and feature information. Firstly, HOG feature extracted from the whole face is selected to represent structural information. Secondly, the feature information is described using a sparse kernel prototype representation method in local face components. The structural and feature information are applied for sketch-photo matching respectively, and then the matching scores are fused for final recognition based on Nearest Neighbor Classifier. The experimental results on different databases based on the proposed method, demonstrate the outperformance of our method compared with state-of-the-art methods.
引用
No.****
同行评议
共计0人参与
勘误表
结构信息和特征信息结合的素描人脸识别
评论
全部评论0/1000