Visually asymmetric image encryption algorithm based on SHA-3 and compressive sensing by embedding encrypted image

This paper presents a new asymmetric image encryption and hiding algorithm based on SHA-3 and compressive sensing. First, to resist the chosen-plaintext attack (CPA) and the known-plaintext attack (KPA), SHA-3 is employed to calculate the hash values of the plain image, and they are transformed into the initial values of the chaotic map. Second, the plain image is divided into blocks which are processed by sparse transform, scrambling, compressive measurement, and merged. Then they are quantized and scrambled again to get an encrypted image. Third, the encrypted image is filled with zeros to the same size as the plain image. Then three numbers from the hundreds digit, tens digit, and single digit of each pixel are extracted respectively. Finally, the carrier image is transformed by integer wavelet transform (IWT) to get four coefficients, and then the above three numbers are embedded into low-frequency coefficients respectively. By inverse IWT, the carrier image containing secrets (CICS) can be obtained. The experimental results show that the algorithm has strong imperceptibility and key sensitivity. Especially, when the embedding coefficient is set to 1.0, the NC value between the carrier image (CI) and the CICS can reach 0.9999, and the PSNR value can reach about 43 dB.