Asymmetric encryption of multi-image based on compressed sensing and feature fusion with high quality image reconstruction

In this paper, an asymmetric multi-image encryption method based on compressed sensing (CS) and feature fusion is proposed. The spectrums of four grayscale images are obtained by optical wavelet transform (OWT) respectively. According to the different features of the wavelet spectral coefficient, the high-frequency and low-frequency components of four images are merged into a high-frequency fusion image (HFFI) and a low-frequency fusion image (LFFI) respectively, and then the HFFI is divided into two matrices after being measured by CS. Subsequently, the two matrices and the LFFI are encrypted to a single ciphertext by phase truncation (PT) and phase reservation (PR) in the Fresnel domain. This proposed scheme not only can successfully encrypt multi-image, but also can recover the primary image with high quality because of avoiding the loss of high-frequency information. Numerical simulations have demonstrated the security and effectiveness of the proposed method.