Research On Iris Image Encryption Based On Deep Learning

In cryptography,the key is important to successful implementation of encryption and decryption.The traditional image encryption algorithm can not resist the attack of malicious key sharing and denial.If the key length is too large,it will be easy to be lost and the artificial memory is difficult.Biometrics-based encryption technology has arisen to solve the problem of poor security of the key.According to the individual biological characteristics,the key is generated,and then applied to the corresponding image encryption algorithm,to achieve information encryption.The iris not only satisfies the above requirements,but also has rich feature information,strong anti-attack ability and excellent encryption potential.Iris image encryption has become an important branch of image encryption field and plays an important role in image encryption.In this paper,Deep Learning algorithm is used to extract the features of the iris images.Based on the original iris image encryption algorithm,a novel iris image encryption algorithm based on deep learning is proposed.The training of the samples,not only realizes the image encryption function,but also decreases the work with high precision.It is a breakthrough to achieve without the key transmission.You can achieve encryption and decryption functions.The main work includes the following aspects:The iris image encryption and Deep Learning are analyzed in depth respectively.This paper introduces the basic principles of image encryption,iris feature extraction,Deep Learning and related techniques.This paper introduces the principle of iris encryption and the problems that need to be solved,and designs the experiment to realize the iris encryption and decryption functions.The iris feature extraction algorithm based on Deep Learning is described,and the experiment is realized.Based on the model of Deep Learning,feature extraction of iris image is carried out.The extracted features are encrypted,and the proposed algorithm is evaluated objectively.The proposed algorithm is combined with the traditional image encryption algorithm and attempts to propose a new biometric-based image encryption algorithm.Based on those,a new keyless image encryption algorithm is proposed for image encryption key transmission security.The Deep Learning model is used to realize the iris features extraction to generate the key,and the encryption and decryption process are realized without key's transmission,and the image encryption security has been greatly improved.