Image Encryption Method For Low Precision Environment

With the rapid development of low-performance multimedia equipment and Internet of things system,there is a great demand for safe and efficient image encryption technology in the environment of low computing accuracy.The mainstream image encryption methods based on chaotic system are often not only unsatisfactory in speed,but also unable to guarantee the security in the low-precision computing environment,and some can not even operate normally,so people’s needs can not be met.In order to solving these problems,a single image encryption method and a batch image encryption method independent of chaotic system are proposed in this paper.（1）An image encryption scheme based on deep Boolean convolutional neural network（DBCNN）is proposed.The scheme consists of multiple convolution layers based on Boolean operation and modular activation of neurons.Different from the existing architecture of deep convolutional neural networks（DCNN）,the proposed DBCNN is only stacked by multiple convolution layers based on Boolean operation,and there is no convolution layer,aggregation layer or variable optimization process based on arithmetic operation that has been used in the existing DCNN.At the same time,unlike the current encryption system based on chaos,DBCNN operates in a low precision(2^-8)environment and has dozens of keys.These keys are generated by a pseudo-random number generator called prime modulo multiplication Linear Congruence Generator（PMMLCG）,which works under the condition of low precision(2^-8),A large number of pseudo-random numbers generated based on chaotic system generally run in or higher precision environment.In addition,the confusion and diffusion processes are combined into one stage to improve the operation speed.The simulation results show that the encryption scheme has high security,low time complexity,strong resistance to common attacks,and the maximum calculation accuracy is.In contrast,the high-dimensional chaos based cryptosystem has low security and cannot withstand some common attacks in some low computational accuracy environments.（2）A batch image encryption method based on prime modular multiplication linear congruence generator（PMMLCG）is proposed.The core of this method is to construct a prime modular multiplication linear congruence generator which can operate effectively in low precision environment,which can not only run under low calculation accuracy,but also produce pseudo-random numbers in a long period,which meets the requirements of image encryption.After that,the image set is divided into three groups,and three combined images are generated;Then the hash function is used to get the hash value of the whole data set,and the third group of combined images is updated according to the formula;Then,the above combined images are generated into an encryption sequence matrix by using the formula;Then all plaintext images in the image set are scrambled and diffused according to the generated sequence matrix,and the XOR operation is used to generate the ciphertext image;Finally,an improved version of 2D-SCL with high security is used to encrypt the encryption sequence matrix.Experiments show that the batch image encryption method proposed in this paper has greatly improved the encryption speed and security under low computational accuracy,but the reproduced contrast encryption method can not meet the requirements.（3）After introducing the two encryption methods in this paper,the fifth chapter introduces the encryption and decryption system based on the above two encryption methods.This system can select single or batch image encryption and decryption according to the needs of users.It is convenient for users to use the corresponding encryption and decryption methods.It can be seen that this method has certain application value.The experimental results show that the two image encryption methods proposed in this paper not only improve the security of encrypted images,but also reduce the time required for encryption,it has certain application value,and also provide a new idea for the follow-up research of efficient and secure image encryption methods.