Reality-preserving image encryption associated with the generalized Hilbert transform

A novel image encryption technique is proposed associated with the generalized Hilbert transform in this paper. Two schemes of this technique are presented in generating real encrypted image and recovering the original image. The encrypted image is real-valued without data expanding, which can benefit the digital processing of images by electronic computers where speed of computation is important. The angles of the 2-D fractional Fourier transform, the phase of the fractional Hilbert transform and the pixel permutation matrix are as keys in the proposed method, which can improve the security of the image encryption. The mean square error between the decrypted and original real image is evaluated; and the robustness with respect to the blind encryption is performed. Simulation results show the validity of the proposed technique.