Interference-based image encryption with silhouette removal by aid of compressive sensing

Compressive sensing (CS) offers the opportunity to reconstruct a signal from its sparse representation, either in the space domain or the transform domain. Exploiting this character, we propose a simple interference-based image encryption method. For encryption, a synthetic image, which contains sparse samples of the original image and the designated values, is analytically separated into two phase only masks (POMs). Consequently, only fragmentary data of the primary image can be directly collected in the traditional decryption scheme. However, the subsequent CS reconstruction will retrieve a high quality image from the fragmentary information. The proposed method has effectively suppressed the silhouette problem. Moreover, it has also some distinct advantages over the previous approaches.