Research On Image Encryption Based On Computational Ghost Imaging

With the rapid development of computer network and all kinds of information transmitting via network,the security of information attracts more and more attention.Kinds of information encryption methods are proposed by scientific researchers,in which the encryption of images attracts great attention.Since Refregier proposed double random phases encoding in 1995.the optical encryption methods develop quickly in information security domain as they have advantages of high capacity,multi-dimension and parallel processing.And the ghost imaging is also applied into image encryption since Pittman proposed ghost imaging in 1995.While the traditional optical image encryption encrypts plaintext image into complex cipher image which contains both amplitude and phase,ghost imaging encryption encrypts image into intensity values.It makes the storage and transmission of the ciphertext more convenient.So researchers devote themselves into the research of information security based on ghost imaging.This paper introduces the theory of ghost imaging systematically,and introduces the image encryption system based on computational ghost imaging in brief.In this paper,on the basis of ghost imaging,several more efficient and secure optical image encryption systems are proposed by combining compressed sensing algorithm,chaotic algorithm and digital holography algorithm.This paper mainly does the following work:(1)We proposed an optical image encryption method with multiple light paths.In the encryption process,M random phase-only masks are generated by means of logistic map algorithm,and theses masks are then uploaded to the spatial light modulator.The collimated laser light is divided into several beams by beam splitters as it passes through the spatial light modulator,and the light beams illuminate the secret images,which are converted into sparse images by discrete wavelet transform beforehand.Thus,the secret images are simultaneously encrypted into intensity vectors by ghost imaging.The distances between the spatial light modulator and secret images vary and can be used as the main keys.According to compressive sensing,we can decrypt the image by orthogonal matching pursuit algorithm.The proposed encryption system decreases the storage of ciphertext,and we do some simulations to verify the feasibility,security and robustness of the system.(2)A computational ghost imaging encryption method based on fingerprint key is proposed.In this work,we first introduce one's fingerprint information into computational ghost imaging for the first time.The phase mask key in computational ghost imaging encryption is generated by a fingerprint image using digital holography method.As the phase key links to one's fingerprint which is uniqueness,this method offers a significant improvement for the security of computational ghost imaging encryption.(3)We proposed a computational ghost imaging encryption based on plaintext-related key.As the linear correlation exists between the reference intensities and the signal intensities in computational ghost imaging,it is possible to chosen plaintext attacks on computational ghost imaging encryption by solving linear equations.To solve this problem,we proposed a computational ghost imaging encryption based on plaintext-related key.In this system,we introduce a plaintext-related key which is obtained by math computing.In the encryption process,we first calculate the sum of the plaintext image gray values,and the sum is used as the seed of linear congruential generator to generate pseudo-random number sequence.The sequence is arranged into a matrix which used as a phase key in computational encryption.As the different plaintexts have different keys,the system can resist chosen-plaintext attack sufficiently.The simulation verifies the feasibility,security and robustness of the system.