Research On Robust Algorithm For Encrypted Face Recognition Based On Neural Network

Face recognition is a hot research topic in image processing,machine vision and artificial intelligence.With the development of science and technology,face recognition systems are widely used in the fields of access control,public monitoring,financial payment,etc.Compared with the traditional methods,such as keys and signatures,it is more convenient and easy to identify identity by face recognition.The cloud platform can be utilized to store face image data.However,the image data stored in the cloud platform may be attacked and stolen,and the data security and privacy issues become more and more obvious.The face is unique as a biological feature,and once it is stolen,it will be hard to modify,so its harmfulness is more than the traditional verification method such as signature.At present,most face recognition methods have no studied encrypted faces,so it is of great significance to study encrypted face recognition.In view of this problem,from the perspective of protecting data privacy and security,this paper mainly conducts the following research:1)The encrypted face recognition algorithm based on neural network and discrete cosine transform is studied to protect the security of original image data.It is mainly divided into three processes:including encryption of face image data,image feature extraction and training and testing of neural network.In the encryption phase,the excellent features of Logistic Map and discrete cosine transform are used to encrypt the image data and protect security of the data.The PCA algorithm is used to preprocess the features of the extracted image and used as the input to the BP neural network for training and testing the neural network.The experimental results show that the algorithm has good recognition rate and robustness.2)The encrypted face recognition algorithm based on neural network and discrete Fourier transform is studied.The algorithm uses a more complex Tent chaotic map to generate keys,enhances the complexity of the key,and further enhances the security of the face image.The image is encrypted in the DFT transform domain,and finally face recognition is completed by combining PCA algorithm and neural network.The experimental results show that the algorithm has better robustness and recognition rate.3)The encrypted face recognition algorithm based on neural network and DWT-DCT is studied to further protect the privacy of image data and improve data security.The algorithm uses two chaotic sequences of high-dimensional chaotic Henon Map and one-dimensional chaotic logistic map to enhance the security of the key,and combines the robustness of the image frequency domain space to further improve the security of the algorithm.The simulation experiment and the algorithm comparison show that the algorithm has good resistance to conventional attacks,geometric attacks and occlusion attacks.