基于DCT的PCA及神经网络的人脸识别研究

PCA(Principal Component Analysis,也称为K-L变换)方法在许多人脸库的识别中取得了很好的效果,但运算量大;DCT(Discrete Cosine Transform)有较高的压缩性能并可通过快速算法来完成;集成的BP神经网络比单一BP网具有更好的泛化能力.针对上述特点,提出了一种新算法,首先用DCT处理原始人脸图像,然后用包含了原始图像大部分信息的少量DCT系数作PCA,最后用集成的BP神经网络进行识别.在ORL人脸库上的仿真试验结果表明,该方法是一种更为高效的识别方法.

Face Recognition Using PCA and Neural Network Based on DCT

PCA(Principal Component Analysis,or K-L Transform) has been widely used in many faces databases with good results,but it requires larger computation.DCT(Discrete Cosine Transform) has better data compression performance and the fast computation ability.an integrated neural network has better generalization ability than single BP neural network.A new human face recognition method is proposed,which deals with the original face images using DCT first,and then applies PCA to a few DCT coefficients that contain most of the information of the original images.At last,an integrated BP neural network is built as a recognizer.The results of simulation experiments based on face images in ORL face database show that the method is more efficient for face recognition.