Research On Face Information Recognition Based On Convolutional Neural Network

Face recognition is an important and popular research topic in computer vision.Convolution neural network algorithm has a unique advantage in face recognition,and its advantage is derived from its weight sharing feature,local perception structure and pooling operation.Closer to the real world of biological neural networks,reducing the complexity of the network.The weight sharing feature can effectively reduce the complexity of the feature extraction and classification,the local perception structure and the pooling operation can reduce the number of computing parameters of the network.These factors determine the image processing problem of the convolution neural network In the advantage.Face recognition is mainly a face image recognition,so face recognition based on convolution neural network is worthy of study.In the face of great ambiguity attitude or image acquisition vary greatly,face recognition accuracy is not high,in order to improve the face recognition rate,this paper proposes a new information enhancement based on convolutional neural network face recognition algorithm of fuzzy face image acquisition wavelet denoising processing,image noise reduction of output adaptive template matching based image segmentation method,the face images are divided into blocks,using geometric feature invariant Radon wavelet transform the key feature points of the face to enhance information,by convolution of facial feature points enhanced classification neural network classifier,feature point extraction and optimization of face accurate identification.The simulation results show that the method of face recognition accuracy is better,and can meet the application requirements for rapid identification of large quantities of samples of face.