摘要
研究基于面部轮廓曲线特征的三维人脸识别。为提取最优面部曲线特征,提出一种基于模糊聚类方法的人脸曲线特征优选算法。该算法从三维人脸深度图中选取最具代表性的8条轮廓曲线,作为主要识别特征,这在很大程度上降低了计算复杂度,克服表情和光照对人脸识别的消极影响,同时它用最少的轮廓线包含了所需要的人脸识别信息。基于这些人脸轮廓线特征,利用改进的Manhattan距离分类器来提高人脸识别的分类效果。实验结果表明,所提出的算法明显提高了人脸识别率,并且具有很强的鲁棒性。
The 3D face recognition method is studied on the basis of facial contour features. To obtain the most important features from the face curves, the algorithm of optimization is proposed by use of the fuzzy cluster method. Eight most representative contour curves are chosen from 3D face depth images to identify the face characteristics, which serves to reduce the complexity of computation, overcome the adverse effects of face expressions and ambient illumination on face recognition and contain all required face recognition information with help of the minimum number of contour lines. The modified Manhattan-distance classifier is used to accomplish face recognition with those chosen curves to improve the classification accuracy. Experimental results show that the proposed methods and algorithms raize the face recognition rate obviously and are of strong robustness.
出处
《铁道学报》
EI
CAS
CSCD
北大核心
2012年第1期34-38,共5页
Journal of the China Railway Society
基金
国家自然科学基金(60973060
61172128)
机器人学国家重点实验室基金(RLO200801)
轨道交通控制与安全国家重点实验室基金(RCS2010K02)
中央高校基本科研业务费专项资金(2011JBM019)
