Due to the problem of security for information encryption system, an image information encryption scheme combined compressed sensing with optical theory was proposed. At the transmitted terminal, image information encryption based on compressive sensing was realized by sparse representation and random projection firstly. Then, the measured values with low data volume after dimensional reduction were re-encrypted by double random-phase encoding technique and then dispersed and embedded into the host image. At the received terminal, original image information was reconstructed approximately via Orthogonal Matching Pursuit algorithm after the inverse process of double random-phase encoding technique. Not only the security of information was ensured under the premise of decreasing the amount of data transmission and reducing the size of the random phase masks but also the privacy of keys were assured which were gained by rules rather than transmitted from transmitter to receiver. Numerical experiments showed that the quality of decryption image was ideal with the corresponding peak signal to noise ratio of 30.899 1 dB and this system had the advantages of good performance of anti-cropping, anti-noising, anti-rotating and anti-filtering, strong robustness and high security.