人脸识别是计算机视觉和模式识别领域的一个活跃课题,有着十分广泛的应用前景。提出一种新的人脸识别方法,该算法采用K-L变换来提取最终特征值,采用了近邻法设计分类器,这样可以有效的避免了利用每一类的“代表点”设计分段线性分类器问题,减小了识别的错误率。在windows XP 平台和MATLAB 7.0 软件环境下,采用上述方法编制了人脸识别系统,仿真结果表明了该方法的有效性。
关键词:人脸识别,特征提取,K-L变换,线性分类器
ABSTRACT
The face recognition is an active subject in fields of computer vision and pattern
recognition, which has a wide range of potential applications. In this paper, a method
face recognition is presented, this algorithm extracts the final features by utilizing the techniques of the simulative K-L transform. The near neighbor method based algorithm was utilized for classification, the problem and difficulty can be effectively avoided, which were faced when piecewise linear classifier was designed using each of the classes of "representative spots" piecewise, and the error rate of the recognition could be reduced. The face recognition system was developed and implemented on Microsoft Windows XP platform and Matlab 7.0 software environment using the above method. The effectiveness of the approach is experimentally demonstrated.
Keywords: Face recognition, Feature extraction, K-L transform, Liner classifer