摘要
人脸受到年龄、表情、化妆品等影响,使人脸识别具有局限性,而人耳正好弥补了人脸的不足,提出了基于正面人脸和人耳相结合的识别方法。利用二维Fisher线性判别方法分别进行了人脸、人耳、人脸人耳组合识别,其中组合识别包括脸耳相互辅助识别、脸耳图像组合识别、脸耳特征按不同权重组合识别。在北京科技大学(USTB)人耳图像库和ORL人脸图像库上进行实验,实验结果表明,人脸人耳组合识别的识别效果好于单一的人脸识别和人耳识别。
Because facial features is unstable under impacts of age,facial expressions,and cosmetics,face recognition has the limitation.And ear physiological characteristics make up exactly the face＇s insufficiency.To solve the problem,the combination recognition method based on the positive face and ear is proposed,to enhance recognition effect.Two-Dimensional Fisher Linear Discriminant is utilized to carry on respectively the face recognition,ear recognition and combination recognition including the face and ear mutually auxiliary recognition,face and ear images combined and facial features and otic features combined according to different weights.Using USTB ear database and ORL face database to test,the result indicates that the combination recognition based on the positive face and ear is better than face recognition and ear recognition.
出处
《微计算机信息》
2010年第25期38-40,共3页
Control & Automation
基金
基金申请人:苑玮琦
项目名称:结合人脸不变特征的人耳识别理论与应用研究
基金颁发部门:辽宁省教育厅创新团队项目(2006T102)
