| In the recent five or six years, digital information hiding technique has become a hot focus in information science field, and it integrates theories and techniques of various subjects. The most remarkable merit of information hiding technique is that any third party, besides the two communicating parties, doesn't know the existence of hidden information, which provides more protection than Encryption and makes the protected information from misunderstanding to invisibility.This dissertation creatively introduced information hiding technique to remote sensing area and developed the connotation of information hiding, namely it was not to hide one of two impertinent information into the other, but to choose information B as the secrete information which was part of or relative to information A and hide it into information A. Remote sensing images possess such characters as large data amount, expensive price, extensive applications and various types such as single-spectrum images, multi-spectrum images, fused images and hyper-spectrum images and include some secrete information and ground objects applied to special purposes, therefore there is much difference between information hiding technique for remote sensing images and the one for general digital images in technical characters and requests, algorithm characters and requests and technical purposes. Furthermore information hiding technique for remote sensing images is a practical one tied up with concrete applications and possesses quite extensive purposes, so this technique holds very important research value and practical value. In the present as for information hiding technique for remote sensing images, foreign researchers are mostly engaged in watermarking technique and copyright protection for remote sensing images and pay little attention to the new technique; and as to our country, the author hasn't consulted any references on the area of information hiding and watermarking techniques for remote sensing images. And as for information hiding technique, in the present national and international researchers are all primarily engaged in research on information complete hiding and there are few reports and references on secrete information hiding technique based on different authorities, which may be due to the need of keeping secrete.This dissertation firstly systemically studied theories, methods and algorithms of the techniques of information hiding, steganography and steganography analysis; analyzed purposes of information hiding technique for remote sensing images and itsdifference from the technique for gerneral digital images; systemically studied and discussed the requirements in system-level and application-level of information hiding technique for remote sensing images; in the domain of information hiding technique for remote sensing images, extended the concept "Invisibility" to "Near-lossless"; in a creative way addressed the new idea of authorized exploitation of remote sensing imags, namely hiding to different degrees secrete information in remote sensing images according to users' different authorities; for the first time addressed an Information Hiding Management System Model (IHMSM) based on different authorities; then lucubrated and realized various information hiding techniques in the area of remote sensing images such as the information hiding technique for single-spectrum remote sensing images, the technique for multi-spectrum and hyper-spectrum remote sensing images, the technique for fused remote sensing images, the technique for various remote sensing images based on authorized exploitation, the technique for various ground objects in one remote sensing image based on different authorizations, the technique for annotation information about the ground objects of some secrete purposes in remote sensing images and the technique for secrete commercial information in remote sensing images; also addressed the new, practical and blind algorithms respectively of spacial domain and wavelet domain and systemically analyzed and compared their performances and applicable circumstances; made a lot of experiments of information hiding for various remote sensing images, the results of which testified the validity, applicability and practicability of the information hiding techniques and algorithms for remote sensing images proposed in this dissertation. And this dissertation also designed and exploited an information hiding archetypal system RSImageHiding for remote sensing images, which included three major functional modules such as information hiding module, application module and attack simulating and testing module.This dissertation mainly includes the following research contents and productions:1 -. This dissertation systemically concluded and summarized fundamental principles and concepts of information hiding, pointed out connection, distinction and compensating relation between information hiding technique and conventional Encryption; addressed the universal model of information hiding system; deeply discussed and studied classification of information hiding technique, designrequirements of information hiding system and information hiding analysis technique; and briefly and deeply expounded primary branches of information science, application of spread spectrum communication technique in information hiding and information theory method of information hiding.2^ This dissertation concretely expounded fundamental principles and concepts of steganography technique; deeply analyzed and discussed classification of steganography technique and structure and security of steganography system.3^ This dissertation reviewed various sorts of steganography techniques in the past years: replacing system, transform domain technique, spread spectrum technique, statistical method, distortion technique and carrier producing method; concretely analyzed and discussed the corresponding algorithms exploited by each sort of steganography technique; discussed many flexible and simple steganography systems in noise communication channels; addressed the merits and shortcomings of each kind of technique and algorithms and also illuminated the idea that in practical applications we should eclectically consider, choose and exploit the suitable technique and algorithm according to the concrete need and requirements.4> The dissertation concluded and summarized fundamental principles and concepts of steganography analysis technique; dissertated classification of steganography analysis technique and steganography analysis attack model; deeply discussed and studied steganography analysis technique based on image information and introduced international typical techniques and algorithms of steganography analysis.5> This dissertation analyzed the purposes and values of information hiding technique for remote sensing images; studied and discussed its difference from the technique for general digital images in characters of images, technical characters and requests, algorithm characters and requests and technical applications; addressed the requirements of the new technique in system-level and application-level; illuminated that the new technique for remote sensing images should satisfy not only the requirements of information hiding technique for multimedia, but also its own particular requirements such as near-lossless, option, blindness and robustness; addressed the new idea that information hiding technique should firstly content the preconditions of imperceptibility, large information hiding capability and almost no influence on various applications of remote sensing images and then secondly consider its robustness; and introduced the concept of "Near-lossless" into the domain of information hiding technique for remote sensing images as supplementation of the concept "Invisibility"; creatively addressed the new idea of authorized exploitation ofremote sensing images, namely hiding to different degrees secrete information in remote sensing images according to users' different authorities; and for the first time addressed an Information Hiding Management System Model (IHMSM) based on authorized exploitation.6> This dissertation systemically lucubrated and realized various information hiding techniques in the area of remote sensing images such as the information hiding technique for single-spectrum remote sensing images, the technique for multi-spectrum and hyper-spectrum remote sensing images, the technique for fused remote sensing images, the technique for various remote sensing images based on authorized exploitation, the technique for various ground objects in one remote sensing image based on different authorizations, the technique for annotation information about the ground objects of some secrete purposes in remote sensing images and the technique for secrete commercial information in remote sensing images; discussed and addressed their technical ideas, technical guidelines and flows and key realized techniques and algorithms.7> This dissertation addressed the new algorithms respectively of spatial domain and wavelet domain and introduced correctness-coding technique and spectrum-spread technique. The algorithms proposed in the dissertation could adaptively hide secrete information according to the features and textural energy of different regions in the remote sensing image and were blind ones which didn't need the original remote sensing images while extracting secrete information and resuming remote sensing images. The dissertation made systematical analysis and comparison of the spatial algorithm and the wavelet one in such 6 aspects as information capacity, accuracy of the exacted information, robustness, introduced correctness-coding, influence on applied value of remote sensing images and algorithmic complexity, and the dissertation also illuminated their own suitable circumstances.8^ This dissertation exploited SPOT5 image, TM1-7 image, SAR image and the fused image of panchromatic one and multi-spectral one to make all kinds of information hiding experiments for remote sensing images, such as the ones for single-spectrum images, for multi-spectrum images, for fused image based on different authorities, for various secrete ground objects in the remote sensing image based on different authorities, for secrete annotation in the remote sensing image and for commercial secrete information in the remote sensing image. And the experiments all verified validity and practicability of various information hiding techniques and algorithms for remote sensing images proposed in the dissertation, which have theadvantages of good transparency, robustness, large information hiding capacity, exactly extracted secrete signal, vividly resuming the remote sensing images, low algorithmic complexity, quite strong robustness against image noise-adding and JPEG lossy compression and almost no influence on applied value of remote sensing images and are practical information hiding techniques and algorithms for remote sensing images.SK This dissertation designed and exploited an archetypal information hiding system RSImageHiding for remote sensing images, which included three functional modules: information hiding module, application module and attacking simulation module of remote sensing images. And the archetypal system realized such functions as information hiding for various kinds of remote sensing images, including single-spectral remote sensing images, fused remote sensing images and multi-spectral remote sensing images, single or multiple secretes hiding in one remote sensing image based on different authorities, secrete annotation hiding in remote sensing images, the two spatial information hiding algorithms for remote sensing images, the embedding-areas of which were respectively the pixel-value compensation area and the complicated textural area that users didn't take interest in, and the wavelet algorithm based on HVS character. The archetypal system RSImageHiding also realized the statistical evaluation for the performance of the information hiding techniques and algorithms of remote sensing images, realized various applications of remote sensing images, such as edge detection and its statistic characters computation and realized attacking simulation for remote sensing images.10 > This dissertation illuminated that at present the information hiding technique for remote sensing images was still at the underway moment, put forward the existed problems and further research of the technique in theory, technique and method and advanced further improved aspects of the technique and algorithm proposed in the dissertation.The dissertation has the following primary creations:1-. This dissertation systemically analyzed, concluded and advanced the value and significance of information hiding technique for remote sensing images, its difference from and homology with the technique for general digital images and its requirements in system-level and application-level; proposed the new idea that the technique should firstly content the preconditions of imperceptibility, large information capability and almost no influence on various applications of remote sensing images and thensecondly consider its robustness; and introduced the concept of "Near-lossless" into the new technique domain as supplementation of the concept "Invisibility". At present worldwide researchers are all most engaged in the research of information complete hiding and concern little about the technique of information authorized hiding, and the dissertation creatively addressed the new idea of authorized exploitation of remote sensing images and for the first time advanced an Information Hiding Management System Model (IHMSM) based on authorized exploitation of remote sensing images.2^ This dissertation creatively introduced information hiding technique into remote sensing region. At present international researchers are most pursuing the research of watermarking algorithm and copyright protection of remote sensing images and pay little attention to the technique of secrete information hiding in remote sensing images. This dissertation lucubrated and realized various information hiding techniques for remote sensing images such as the technique for single-spectrum remote sensing images, the technique for multi-spectrum and hyper-spectrum remote sensing images, the technique for fused remote sensing images, the technique for various remote sensing images based on authorized exploitation, the technique for various ground objects in one remote sensing image based on different authorizations, the technique for annotation information about the ground objects of some secrete purposes in remote sensing images and the technique for secrete commercial information in remote sensing images.3> This dissertation addressed 2 algorithms in spatial domain for information hiding in remote sensing images and exploited different embedding strategies according to different embedding regions. The 2 spatial algorithms are both blind ones, could adaptively choose quantization factors in terms of regional features and textural energy of remote sensing images. In addition the dissertation also introduced correctness-coding technique and spectrum-spread technique to further improve robustness of the 2 spatial algorithms and in-depth compared, analyzed and illuminated the advantages, disadvantages and suitable applied circumstances of the 2 algorithms. And the experimental results show that the 2 spatial algorithms have the advantages of good information hiding effect, large information hiding capacity, exactly extracted secrete signal, vividly resuming the remote sensing images, low algorithmic complexity, no influence on applied value of remote sensing images, much strong robustness against image noise-adding and strongish robustness against JPEG lossy compression, and are effective and practical information hidingalgorithms for remote sensing images.4> This dissertation addressed a wavelet algorithm for information hiding in remote sensing images. The dissertation introduced information embedding strategy of DWT domain and adaptive embedding strategy based on HVS into the wavelet algorithm and could adaptively embed the secrete information by computing the just noticeable distortion threshold JNDT of the middle-low frequency wavelet coefficients and their textural characters and orders. Furthermore the dissertation also exploited the idea of secrete information regroupment that the important bits of secrete information should embed into those important wavelet coefficients and unimportant bits into unimportant wavelet coefficients; and introduced correctness-coding technique and spectrum-spread technique to further improve the accuracy of the extracted information and the robustness of the wavelet algorithm. The wavelet algorithm is also a blind one which doesn't need the original remote sensing images while extracting the secret information and resuming the remote sensing images; and has the advantages of good transparency, large information hiding capacity, exactly extracted secrete signal, vividly resuming the remote sensing images, almost no influence on various applications of remote sensing images, quite strong robustness against JPEG lossy compression and strongish robustness against image noise-adding, and is a practical information hiding algorithm suitable to remote sensing images. The dissertation made systemic and deeply comparison and illumination of the characters, advantages, disadvantages and each own suitable circumstances of the spatial algorithm and the wavelet one proposed in it. And the experiments testified the spatial algorithm and the wavelet one proposed in the dissertation can both well preserve the textural and spectral features of the secrete ground objects and vividly resume the original appearance of the remote sensing images.5> This dissertation developed the connotation of information hiding, namely it is not to hide one of two impertinent information into the other, but to choose information B which is part of or relative to information A as the secrete information and hide it still into information A. As for the present most information hiding algorithms, the secrete information to embed is irrelative to the carrier. So the algorithms proposed in the dissertation should consider not only invisibility and robustness of secrete information B but also damaged degree and applied merits of carrier information A after being embedded secrete information and suffering from attacks. Unlike the general information hiding algorithms, this algorithms cannotreject the carrier information after extracting the secrete one.6x In the present information hiding techniques, secrete information is generally a binary sequence or a two-value image and there haven't still reported algorithms or papers with a colorful image as the secrete information. This dissertation studied and realized the hiding of secrete gray images or colorful ones7^ At present information hiding techniques all focus on the research of single image, however the dissertation realized the information hiding technique for multi-spectrum, even hyper-spectrum remote sensing images.8^ This dissertation proposed 2 production algorithms of the disguised remote sensing images: Gauss Filtering method and Image Fusion method.9> The dissertation exploited the idea of image delamination and classification to realize the decomposition, analysis and integration of secrete ground objects.
|
PDF Full Text Request