基于DCT和BP神经网络集成的人脸识别

Face recognition based on DCT and integrated BP neural network

人脸识别是模式识别和图像处理领域的研究热点和难点,尽管已提出了许多方法,然而如何在变化的环境下实时、高效地识别人脸仍是一个难题.鉴于离散余弦变换(discrete cosine transform,DCT)有较高的压缩性能和快速算法以及集成神经网络良好的泛化能力,提出了基于DCT和神经网络集成的人脸识别方法.首先用DCT提取人脸特征矢量,然后构建集成BP(back propagation)神经网络对人脸进行分类识别.在ORL(olivertti research laboratory)人脸库上的仿真实验结果表明,提出的方法取得了较快的训练和识别速度、较高的识别率,因此该方法是一种快速高效的人脸识别方法.

Human face recognition is a focus and nodus in pattern recognition and image processing areas, numerous approaches have been reported. However, it is still a difficult task for a machine to recognize human faces in real-time under variable circumstances. Based on better data compression performance and the fast computation ability of DCT（discrete cosine transform） as well as the better generalization ability of integrated neural network, in the paper, a human face recognition method is proposed. First, eigenvectors of the original face images are extracted by using DCT. And then an integrated BP （ back propagation） neural network is constructed as a recognizer. Simulation experiments are conducted based on face images in ORL（ olivertti research laboratory）face database. The results show that the method achieves high training and recognition speed, high recognition rate, so the method is efficient for face recognition.