Research On Image Encryption And Authentication Algorithm Based On Fourier Single-pixel Imaging

As we all know,information is easily stolen and tampered by criminals when it is transmitted in plaintext.As a common information carrier,image encryption has become an important branch in the field of information security.Compared with traditional imaging,single-pixel imaging(SPI)technology has great advantages in the noise and low light environment.Fourier single-pixel imaging(FSI),as a typical SPI technology,can image with high quality.However,there are still disadvantages in SPI applications,such as the longer measurement time and the larger number of measurements.According to the above problems,image authentication and encryption schemes are proposed based on the combination of FSI,optical processing technology,and logical map in this thesis.The main contents are as follows:1.A multi-image authentication scheme is proposed based on FSI.Initially,the low-frequency Fourier basis patterns that has up-sampled and binarized,uniformly and randomly illuminates multiple images to be encrypted.In order to ensure that no single-image information can be extracted during authentication,the projection sequence is randomly scrambled and the intensity values collected by the single-pixel detector.Finally,the intensity value sequence is scrambled by the chaotic sequence generated with the logical map and used as the ciphertext.During the decryption and authentication,the nonlinear correlation between the recovered mixed image and the down-sampled plaintext image is calculated to determine the existence of the plaintext.Numerical simulations show that the proposed method achieves image authentication at an average single-image sampling rate of 1.33%and can effectively resist the noise attacks and occlusion attacks.2.An image encryption scheme is proposed based on FSI and phase iteration.Firstly,the bitwise exclusive-OR(XOR)operation is performed between the plaintext image and the random matrix generated by the logical map.Then,the XORed image is encoded into the target image,which generated by the blended binary barcode,by using the phase iterative process in Fresnel transform domain.Finally,the two-step phase-shift FSI(TP-FSI)is performed on the binary barcode,and the collected intensity value is mixed with the random intensity value of the same magnitude and scrambled as the ciphertext.The proposed method can effectively reduce the number of measurements for SPI.It is worth noting that the proposed method can losslessly decrypt image when the ciphertext is subjected to a low-intensity occlusion attack or noise attack.The feasibility,safety,and robustness of the proposed method have been verified with a large number of numerical simulations.3.A color double image encryption scheme is proposed based on FSI and double random phase encoding(DRPE).The original color plaintext images are encrypted into two gray-scale intermediate ciphertexts with the method of the encrypted color Fourier basis patterns and FSI.Two new images are obtained by performing the splicing-scrambling-decomposition processes on the intermediate ciphertexts with the help of the chaotic sequence generated by the logical map,one is modulated by the phase-only mask,which generated by the means of performing the phase iteration process on the other image,and then encrypted as the real-valued ciphertext by DRPE.The gray-scale real-valued ciphertexts reduce the number of ciphertexts and hide color information.With the aid of a linearly independent coefficient matrix,the phase keys of the output and transformation plane is recalculated,which solves the problem of vulnerability when encrypting images using DRPE.At the same time,the proposed method has passed the robustness tests such as occlusion attacks,noise attacks,and selected plaintext attacks.