An asymmetric image encryption scheme based on hash SHA-3, RSA and compressive sensing

Objective: In order to avoid the unauthorized access to private images and make a secure communication, an asymmetric image encryption algorithm based on SHA-3 and compressive sensing is proposed.Methods: Firstly, a random matrix is generated and a modular-addition operation is performed between it with the plain image, obtaining a preprocessed image. Then, hash values of the pre-processed image are computed by secure hash algorithm in the third generation (SHA-3), and are grouped and summed to get three plaintext keys. Using Rivest-Shamir-Adleman (RSA), three ciphertext keys can be obtained correspondingly. Secondly, a new mathematical transformation model (MTM) is designed to transform all keys into initial values for chaotic system. Then, keystream is calculated accordingly. Thirdly, the plain image is compressed by compressive sensing (CS), and further confused by random sequences. Then, applying discrete wavelet transformation (DWT) to the confused image and generating four components of high and low frequencies. Chaotic sequences are used again to confuse the low frequency components, and then all components are recombined again into a matrix. Thereafter, perform inverse DWT (IDWT) to get a middle cipher image (MCI). Finally, another random matrix is generated by chaotic se-quences, and the final cipher image is obtained by a modular-addition operation again to the MCI. Results: Experiments show that the proposed scheme can reconstruct the plain image with a high quality. In particular, the proposed method can resist known plaintext attack and chosen plaintext attack because the keystreams are related to the plain image and with no extra transmission.