An Improved Separable and Reversible Steganography in Encrypted Grayscale Images

Reversible data hiding in encrypted images (RDH-EI) has attracted various communities that deal with security. Xiao et al. presented RDH-EI that uses pixel value ordering (PVO) and additive homomorphism for encryption ensuring that the embedding capacity of plain and encrypted image remains relatively same. However, the size of location map required to manage over/under flow blocks increased drastically reducing the overall embedding capacity. In the paper, a new PVO-based separable RDH scheme is proposed using double encryption. The original image is encrypted using chaotic mapping in Phase 1, ensuring that the performance of PVO algorithm in an encrypted domain is similar to that in plain domain. Secret data is embedded in this encrypted image. The embedded image is re-encrypted using additive homomorphism to enhance security. The proposed scheme is more efficient as it improves the pure embedding rate and total embedding capacity of the image by reducing location map size, overcoming the drawback of Xiao et al.