摘要
提出了将自适应主分量提取神经网络(APCENN)与径向基神经网络(RBFNN)结合进行人脸识别的方法.由于人脸图像维数高,传统主分量分析方法提取人脸主分量运算复杂、速度慢,应用APCENN通过并行运算直接提取人脸主分量,提高了特征提取速度.再通过RBFNN进行识别分类,实验证明网络训练收敛速度快、识别率高.
A method of integrating （APCENN） and （RBFNN） for face recognition is presented. The number of dimensions of face image vector is large, and the algorithm which uses traditional principal component analysis to extract the principal component of face image is slow and complicate. By taking advantage of parallel calculation of APCENN to directly extract principal component, the speed is improved. The output of APCENN is sent to RBFNN for face recognition and classification. Simulations show that training network is fast and the method can reach a high recognition rate.
出处
《兰州交通大学学报》
CAS
2007年第1期15-17,共3页
Journal of Lanzhou Jiaotong University
基金
教育部高等学校博士学科点专项科研基金(20060732002)
甘肃省自然科学基金项目(ZS031-A25-019-G)
