Face Recognition Based On Wavelet Transform And Pulse Coupled Neural Network

In recent years,with the deepening of the development of network information society,people pay more and more attention to personal information security,a variety of identity information security verification methods emerge in an endless stream,and face as an effective identification feature has always been a scientific researcher focus.The research on face recognition is not only important theoretical value,but also has great practical application value.The main research of this paper is to use wavelet transform and pulse coupled neural network to extract the feature of face.Wavelet transform is known as "mathematical microscope",and its appearance is a perfect solution to the shortcomings of Fourier transform in signal processing,and has been widely used in signal analysis,image recognition and machine vision.In this paper,we first use the wavelet transform to preprocess the face image to extract the low frequency sub-image of the face image.This low-frequency sub-image contains the main information of the human face,which achieves the dimensionality of the data and blurs the facial expression and attitude And so on.The pulse-coupled neural network is a new artificial neural network model based on the iterative,automatic monitoring and self-learning based on the mammalian vision system,so it has a unique advantage in image processing.In this paper,we use the time series of pulse coupled neural network as the recognition feature of human face,and use Euclidean distance to measure the similarity of face to complete face recognition.This method has good robustness to the geometric transformation of face image.In this paper,the traditional single-channel pulse coupled neural network model is extended to a multi-channel model to extract the characteristics of color face images.The process is to convert the color image from the RGB color space to the HIS color space,and then extract the HIS channel components.The wavelet transform is used to preprocess the low frequency sub-image.Then,it is sent to the multi-channel pulse-coupled neural network.Sequence,and finally combined with the European distance to complete face recognition.The experimental results show that the method is less effective than the single channel model,and has better recognition effect.