Optical Image Encryption And Validation Based On Phase Retrieval Algorithms

Information plays a very important role in our daily life, hence it is imperative todevelop security techniques for protecting confidentiality and identifying authenticity ofprivate information. Compared to the traditional information security technology basedon mathematics and computer, optical information security technology has a natural abil-ity of high-speed and multi-dimensional data processing, which can respond quickly tosecurity requirements with large data, and it has attracted much attention in the filed ofinformation security. In this dissertation, several optical image encryption and validationmethods are proposed based on designing optical retrieval problem models and solvingthem, in addition, cryptanalysis of some optical security systems is also performed. Nu-merical simulations are carried out for demonstrations, and necessary optical experimentsare also presented in support of several proposals. The main works are listed as follows.An optical encryption scheme with asymmetric keys using random binary phasemodulations is proposed. To our knowledge, most of the reported optical encryptionschemes belong to symmetric key cryptosystems, which use the same or similar keys forencryption and decryption. However, it has been demonstrated that such symmetric en-coding structures would be vulnerable to chosen/known plaintext attack if attackers haveaccess to the encryption implementations, and asymmetric encoding can greatly enhancethe security against the existing attacks. The popular optical asymmetric cryptosystemis based on phase-truncation, unfortunately, it can be deciphered by a specific iterativeretrieval algorithm. We propose a novel asymmetric optical image encryption schemethat uses a amplitude-phase mixture retrieval type of Yang-Gu algorithm to realize theone-way binary phase modulations. The private keys generated in the encryption processare randomly distributed binary phases with 0 and π, which are used for generating thedecryption keys by performing the one-way binary phase modulations. In a further dis-cussion of optical asymmetric encoding, we point out that novel asymmetric schemes,algorithms and architectures must be investigated according to the special features of theoptical systems, instead of be strictly limited by the notions of digital asymmetric cryp-tography in current research stage, because mathematical puzzles are hard to opticallyimplement.An optical asymmetric multiple-image encryption scheme using phase multiplexingis reported and demonstrated. The original secret images with same sizes are multiplexedand encoded into a real-valued random distributed noise-like image with equal size inFresnel domain by using modified Yang-Gu phase retrieval algorithm and asymmetricencoding. The encryption is implemented with only one public encryption key, but eachsecret image generates a unique independent phase key for decryption, hence the secretimages can be recovered with the corresponding keys in decryption process. As onlyone ciphertext image and several binary key images with distributions of 0 and 1 arerequired for decryption, the transmission data is apparently decreased. Compared withthe multiple-image encryption schemes based on wavelength/position multiplexing, ourmethod achieves a larger multiplexing capacity and higher security against the existingattacks, and it can be straightforwardly applied into multi-authentication due to its simpleoptical implementation.Besides multiple-image encryption schemes can optimize secure transmission, a si-multaneous image compression and encryption scheme using error-reduction phase re-trieval algorithm is presented. The secret images are simultaneously compressed andencrypted into a real-valued random distributed noise-like image by designing a error-reduction phase retrieval algorithm using the pre-set physical constraints for image com-pression. A predefined compression rate can be achieved by controlling the algorithmparameter. The decryption process is similar to the conventional double random phaseencoding; two random phase keys generated in the compression and encryption processserve as the decryption keys. The original images can be recovered by putting the correctphase keys in the cascaded diffraction planes.A secure optical information encryption and validation scheme using classical Michel-son interferometer is introduced. Two beams’ interference based encoding has been wide-ly used for optical image encryption and validation, which can analytically encode oneimage into two pure phases. The original image can be outputted only when both ofthe phases are correct for interference. However, many pairs of phases can output thesame image after interference, and arbitrary of them would imply the image after diffrac-tion. That bring security vulnerabilities. A specific iterative phase retrieval algorithm isdeveloped by redesigning the two random phases which can output two images after inter-ference, to enhance the security. The novel interference encoding method can be securelyutilize for image encryption and validation, and the decryption and validation process canbe implemented using a classical Michelson interferometer.A secure optical multi-user verification system using dual phase-only correlationis introduced. Generally, image validation can be completed by checking the content(such as interference encoding) or correlation of images. The conventional optical iden-tity authentication schemes are mainly developed from correlation recognition methodscombined with random phase encoding. However, the cryptanalysis attacks, such as cho-sen and known plaintext attacks, are carried out to demonstrate that the classical opticalidentity authentication systems suffer counterfeit and collision problems, due to the con-sistent relation between the images. We propose a novel correlation algorithm betweentwo mutually independent elements(phase and phase, phase and amplitude) using Yang-Gu phase retrieval algorithm,and accordingly develop a secure identity authentication.The proposed method has a high security against the existing cryptanalysis attacks andcan recognize multiple images without collision, which is very suitable for multi-userapplications.