Search

Recently, it has been found that cloud storage still has security risks, and research on the security and privacy of user data and information is still in the early stage. This paper studies the security risks of cloud data, and designs an image encryption scheme based on neural networks. First, the existing neural network model is improved to obtain a new bidirectional activation (BA) neural network, to establish a many-to-one mapping relationship between the key and the chaotic initial value, to hide the original key of the cloud encryption system, and to improve the security and randomness of the key system. Then, a medical image encryption scheme based on dynamic index scrambling and the M-semitensor product diffusion is proposed. Dynamic index scrambling is more flexible than the traditional approach, and its security and efficiency are improved. The diffusion algorithm adopts the semi tensor product operation, and one of the product matrices is composed of a unitary matrix after Schur decomposition of a plaintext image to effectively resist a selective plaintext attack. Performance analysis shows that the encryption algorithm has high security. [ABSTRACT FROM AUTHOR]

Recently, it has been found that cloud storage still has security risks, and research on the security and privacy of user data and information is still in the early stage. This paper studies the security risks of cloud data, and designs an image encryption scheme based on neural networks. First, the existing neural network model is improved to obtain a new bidirectional activation (BA) neural network, to establish a many-to-one mapping relationship between the key and the chaotic initial value, to hide the original key of the cloud encryption system, and to improve the security and randomness of the key system. Then, a medical image encryption scheme based on dynamic index scrambling and the M-semitensor product diffusion is proposed. Dynamic index scrambling is more flexible than the traditional approach, and its security and efficiency are improved. The diffusion algorithm adopts the semi tensor product operation, and one of the product matrices is composed of a unitary matrix after Schur decomposition of a plaintext image to effectively resist a selective plaintext attack. Performance analysis shows that the encryption algorithm has high security. [ABSTRACT FROM AUTHOR]

Biometric technology is increasingly being used in various areas, including government law enforcement, exit and entry management, visitor access control, user identification, and daily attendance management for companies. This technology simplifies the procedure of user identity verification and reduces the time required for verification. However, the security of biological data stored in the system can be overlooked. To address this, we propose a secure biometric feature matching framework that encrypts all information stored in the biometric system. The system decrypts and extracts feature values only when recognition and matching are required, enabling secure matching judgment. This framework protects private data in the biometric system from illegal theft, tampering, and destruction, and promotes research and practical application of biometric security technology. To fully ensure system data security, we propose an adaptive rotation diffusion image encryption algorithm based on double chaos. The algorithm uses double chaos to control image scrambling and leverages the random characteristics of chaos. The algorithm adopts six different rotation diffusion methods, and different plaintext images lead to completely different diffusion methods, making the algorithm adaptive and effective against selected plaintext attacks. Experimental simulations and performance analysis confirm the feasibility and superiority of the proposed algorithm. [Display omitted] • Build a secure biometric data protection framework. • Establish matching relationship between plaintext and ciphertext for biometric data. • Propose a plaintext related adaptive rotation diffusion encryption method. [ABSTRACT FROM AUTHOR]