Font Size: a A A

Research Of Full Color Single-pixel Imaging And Its Application In Optical Information Security

Posted on:2023-04-16Degree:MasterType:Thesis
Country:ChinaCandidate:G QuFull Text:PDF
GTID:2568306617970449Subject:Optical engineering
Abstract/Summary:
The 21st century is an era of informationization.The high-tech industries represented by computer and Internet techniques have developed rapidly.Mobile communication,multimedia,automation,and artificial intelligence have also greatly enriched and facilitated people’s daily lives.In this highly informationized society,information security has also received widespread attention.Images and videos are important media for information transmission in modern life and how to ensure the security during their dissemination and prevent them from being illegally stolen or altered are important issues to be considered.Optical information security methods are one of the emerging techniques developed in recent decades,which utilize the multi-dimensional,high-parallel,high-speed,and large-capacity characteristics of optical systems to perform image encoding,encryption,information hiding,and authentication and so on.Based on full color single-pixel imaging technique,this paper presents a series of new methods of optical information security technique by combining with cryptanalysis,compressed encoding,information hiding,deep learning and other techniques.The specific contents are as follows:(1)An optical color image encryption method based on Henon chaotic system and channels-encoding-based single-pixel detection is proposed.First,the projected patterns are modulated by a specific key K1 and three sets of color pattern sequences are obtained.The random key table is generated by the Henon chaotic system with parameters key K2,which is then utilized to scramble the color projected patterns.Then the detection and encoding process of the target object are completed by the single-pixel system.Last,the light intensity signal is further scrambled with the specific key K1,that is,the final cipher-text image is obtained after Arnold transformation.The Henon chaotic system uses a little key space to generate an encryption table of required size,and the projected patterns modulated by a specific key ensure the multi-channel encoding of the target object during the detection process and improve the security of the encryption system.(2)An optical information hiding technique based on Color Filter Array(CFA)-modulated single-pixel imaging,singular value decomposition(SVD)and redistributed invariant lifting wavelet transformation(RILWT)is proposed.First,the grayscale projected patterns are modulated by CFA,the space and spectrum of the target scene are encoded and detected by single-pixel system.The detected signal obtained by single-pixel detector is then decomposed by singular value decomposition to get its main component as the information to be hidden,which is then embedded into the low frequency sub-image of the redistributed invariant lifting wavelet domain of the host image instead of the overall signal.The introduction of CFA in single-pixel imaging greatly reduce the sampling times and occupancy of hidden information for color imaging system,while singular value decomposition of hiding image and the specially designed wavelet decomposition of the redistribution invariant domain ensure this information hiding technique with higher security and additional robustness against geometric distortion.So the proposed color information hiding method has the advantages of low storage footprint,high robustness and security.(3)An image encryption method based on single-pixel color imaging with deep learning and complex chaotic system is proposed.On the basis of CFA-modulated full color single-pixel color imaging encryption method,a specially arranged ’Cake-cutting’Hadamard matrix is introduced and scrambled by a complex chaotic system to obtain a ciphertext image at a very low sampling ratio.During decryption,the color image is reconstructed through deep neural network from the initially restored grayscale mosaic image of single pixel detection.The ’cake-cutting’ Hadamard matrix is the rearrangement of the original Hadamard matrix,and its combination with deep learning realizes the decoding and reconstruction of full color image with single-pixel imaging system under lower sampling ratios,which can reduce the ciphertext capacity and improve the quality of reconstructed images.The introduction of the complex chaotic system effectively improves the security of the encryption system without increasing the key space.
Keywords/Search Tags:Full color single-pixel imaging, Optical information security, Chaotic mapping system, Deep learning
Related items