Crypt analysis of an image compression-encryption algorithm and a modified scheme using compressive sensing

Compressive sensing acquires image in the form of linear measurements. Utilizing compressive sensing in image encryption makes most of the algorithm vulnerable, due to the linearity property of compressive sensing retained in the encryption process. In this paper, a compression-encryption algorithm based on compressive sensing is analysed. It is demonstrated that chosen plain-text attack can be performed successfully for the scheme mentioned above. This paper shows that the encryption combined with compression adapted in the scheme under analysis can be represented as a single linear transformation. The corresponding matrix of transformation can be computed using the chosen plain-text attack. This matrix enables to encrypt or decrypt any image without knowing the keys. A modified scheme is proposed in which the encryption process is designed to change dynamically depending on the values of the plain image. The dynamic nature of the scheme makes it nonlinear and hence the scheme is secure against the chosen plain text attack.