Research On New Methods Of Robust Optical Encryption And Adaptive Blind Watermarking For Color Image

Optical image encryption and image watermarking technology have attracted more researchers’ attention in the field of information security in recent years,and the Double Random Phase Encoding(DRPE)has become one of the research hotspots in the filed of optical image encryption.In the DRPE-based cryptosystems and image watermarking technologies,the problems related to single-channel cryptosystem,selective cryptosystem,asymmetric cryptosystem,multi-watermark embedding and blind watermarking of color images still need to be solved.This thesis studies and proposes a series of new ideas and methods.The primary works and innovations are listed as follows:(1)An optical single-channel cryptosystem based on the discrete wavelet transform for color image is proposed.In this cryptosystem,the color channels are taken as different detail parts of the wavelet transform,and synthesized into a real-valued grayscale image by the inverse wavelet transform simply and quickly.Then the grayscale image is modulated with the random phase mask.The final encryption result can be obtained by the optical transform.In addition,the proposed cryptosystem is not only suitable for encrypting CMYK color images directly,but also for encrypting RGB color images.(2)An optical selective cryptosystem based on the FRFCM algorithm and face biometric for color medical images is proposed.The cryptosystem firstly uses the FRFCM algorithm to segment the medical image to obtain the region of interest(ROI)image.The random phase mask is generated based on the face biometric.Then the ROI image is modulated with the generated phase mask.The final encryption result can be obtained by the optical transform.(3)An optical asymmetric cryptosystem based on QZ modulation is proposed.In this method,the input image and phase mask are taken as two inputs of the QZ algorithm.The outputs of the QZ algorithm include four parts: two upper quasitriangular matrices and two unitary matrices.The two unitary matrices are regarded as private keys only for decryption.The the upper quasitriangular matrice corresponding to the input image is encoded in the optical transform domain to obtain the final encryption result.The two phase masks are not required in the whole decryption process.In addition,the optical single-channel cryptosystem based on inverse wavelet transform proposed in this thesis is also suitable for the QZ modulation method to realize asymmetric single-channel encryption for color images.(4)An adaptive color multi-watermarking scheme based on hybrid particle swarm optimization and grey wolf optimizer(PSO-GWO)algorithm is proposed.The method first determines the optimal embedding regions by the contrast function of different blocks,then we define a single objective function and utilize the PSO-GWO algorithm to optimize the objective function.Finally,multiple color watermarks are simultaneously inserted into the selected regions of the color host image by discrete wavelet transform(DWT)and singular value decomposition(SVD).(5)A blind watermarking method based on QZ algorithm is proposed.The method takes the mark and the cover image as two inputs of the QZ algorithm to realize embedding,and obtains two upper quasitriangular matrices and two unitary matrices.Then the three parts of the outputs are synthesized by the vector synthesis and inverse DWT-SVD to obtain the watermarked host image.In the extracting process,neither the original mark nor the cover image is required,the blind extraction can be realized.The proposed watermarking methods are not only suitable for color images,but also for audio.In addition,based on a variety of biometric features,the cryptosystems and watermarking methods proposed in this thesis can also be applied to image and audio authentication.Numerical simulation results are shown to verify the feasibility,robustness and superiority of the proposed cryptosystems and watermarking methods.The research results of this thesis enrich the theoretical basis of the DRPE cryptosystems and watermarking methods.