ih2

Res earch on Information Hiding Theoretic ModelGe Xiuhui Tian Hao(Hebei University of Economic and Business, Shijiazhuang, China)Abstract: With the development of Internet, People have paid more attention on one branch of security-information hiding. The paper study the present information hiding mode, gives new kinds of information hiding secure model. Then the paper analyzes the existed communication model of information hiding system and provides a new model, which improved Waston model. We discuss the characteristic of the model. At the end, the paper gives the broad perspective of the research. Keywords: Information Hiding, Theoretic Model, Perceptual Model, Steganography, Waston Model.1 IntroductionAccording to communication theory and information theory viewpoint, this paper proposed information hiding technique theoretic model. Through establishing information perceptual modelto make up the flaws based on the communications system model. This model use discrete cosine transform to divide the image into block, then estimate change perceptual, at end, merges these estimate results to the estimate perceptual distance individually, original perceptual model, which MSE frequently can make many errors, we proposed the model, which through embedding strength repeatedly test and realization, from the result analysis, this model precision is higher, because each model precision all limits in certain distorted scope. At same time, the model adjust the public embedding strength, as well as through embedding symbol to distort the specific perceptual effect maximization, in this foundation, carried on perceive forming embedding on the perceptual model, thus make information hiding technique fundamental research to achieve a new theory altitude.2 Present Theoretic ModelIn the information hiding related literature, it generally uses Simmons proposed model and this model’s variety, the information hiding theoretical model is the foundation of research the information hide. In early time, information hiding research is directly adds secret information to the carrier information, regards the carrier data as noise, if the decoder understand the noise, then during extract step, covert carrier data subtracts the noise namely get secret information. At present, the information hiding is one kind of covert communication model. We need further study the information hiding system model theory, the inspection theory. This has formed the information hiding fundamental research.2.1 Simmons Theoretic ModelIn 1983, Simmons proposed the first information hiding scene description-prisoner question in view of the covert communication. Supposes Alice and Bob is separately prisoner in different prison room, in order to plot escapes from prison, they need mutually secret communication, but all communication must pass through guard Wendy. Wendy may read all prisoner's letter, and decided whether transmits or does not transmit these letters, and he has rights to carry on the revision to the letter, but certainly does not change letter’s content. In order to making Wendy not suspects the communication, Alice and Bob cannot use the commonly encryption coding communication technology, because after encryption, the letter is semantic chaotic, although possibly cannot reveal the plan, but it become the confidential letter, it already sufficiently took two criminals plot or conspire evidence. Therefore in the2-227 1-4244-1135-1/07/$25.00 ©2007 IEEE.prisoner question, Alice and Bob not only must guarantee the confidential letter cannot explain, moreover must use covert communication in the fact. Prisoner question like fig 1 shows.Fig 1 Prisoner ProblemThis is the typical information hiding questionfigure of symbols. This model emphasize how to guaranteed covert communication’s realization, but it hasn’t certainly proposed the realization information hiding principle, therefore in fact, this model is covert communication model. We usually called the covert information as secret information, the carrier has embedded secret information to be called stego-carrier, information hiding process which generally control by stego-key, through embedding stego-algorithm, which can embedded the secret information in the public information, then through covert channel to transmit the covert carrier. The receiver of communication opposite party uses secret key with detector to restore from covert carrier to extract secret information. The secret information may be the copyright information or the secret data of sequence number and so on; but the public information is called carrier information like video, audio, image and so on.From above analysis, we may know information hiding technology mainly is composed by two parts: information embedding stego-algorithm, it realizes the secret information hiding using the key. The covert information inspection/extraction algorithm (detector), it uses the key to inspect from the hiding carrier/restores the secret information. Under the key unknown premise, it is very difficult to the third party obtains/deletes secret information from covert carrier, even it has discover secret information.3 CommunicationSystemModelFig 2 has introduced a tradition communications system basics structure. M is prepares transmission information, to information M, channel encoder carries on coding, then prepare transmission, it will mapping all possess possible information into symbol, in this set, the latter can be obtain from the coding set through communicationchannel. The symbol sequence usual mark is X. Standard communication model like fig 2 shows.Fig 2 Standard Model of Communication SystemIn order to further guaranteed transmission information security, it may use password technology to improve the standard model, based on key channel encoded communications channel basic model like fig 3 shows.Fig 3 Communication Channel Model Based on SecretKeyBut in the information hiding communication also should consider likes the Wendy such initiative attack existence, therefore basic information hiding model like fig 4 shows:2-228Fig 4 Basis Information Hiding Theoretic ModelIn fig 4, we discusses system conforms to the Kerchoff criterion, namely the encryption function, the decipher function as well as hiding information embed function and extract functionispublic.4 Secure Model AnalysisRegarding information hiding, multilayer safetymechanism can make secret information safer. In information hiding, encrypts, embedding algorithm as well as covert communication is very important. Maintains secret key algorithm is very difficult. Because outstanding cryptographic system follows the K erchoff principle, the algorithm safely is decided by the key, if does not have the security key, any attack can break through embedding algorithm probability to be very low. Therefore, many encryption algorithm is publicizes uses. If text information through unsafe channel, and the text information already passed covert processing, so it certainly could not suspect between communication both sides carried on covert communication. Moreover, if carries on multilayered safe processing to the text; then text information security will be higher. To encrypted information, then use information hiding tool to carry on processing, at the end, using covert channel to transmit the text. Ifattackers examine secret information existence, he needs to extract primary information, like above description, extracting primary information probability to be very low. If attackers extract primary information, but he also need secret key to decipher, then he can obtains primary information.Although he had discovered secret existence, whichcaused information hiding defeat, but this still could protect covert information not to explain.According to the above basis theory, we propose information hiding security model like fig 5 shows:Fig 5 Information Hiding Security Model5 Present Model DefectsInformation hiding theoretical model Based on information theory has important theory significance. Above we discuss each kind of model to accomplish different research instruction function in different times. What is the modern information hiding theory basis? Is the noise theory or measured does not permit the principle? Does the Shannon information theory model whether complete solution interpret information hiding? Has limited researcher's mentality to this kind of question wrong model, has following several important deficiency based on the information theory model:(1) First model emphasize how to guard covert communication realization, it has not certainly proposed realization information hiding principle, therefore in fact, this model is covert communication model.(2)In information theory, entropy is the solution indefinite question data tool, but certainly cannot determine each image’s entropy, therefore entropy certainly cannot examine in the image whether contains secret information, therefore this model is only theoretically model, in actual realization, has many defects.(3) Majority of model don’t make quantitative analysis to robustness. At present majority of2-229models all strength to hiding volumetric analysis, but has not been able to robustness, the hiding and secure and so on important target carries on the quantitative analysis.In order to solve information hiding’s actual problem which in the model exists, this paper according to states the knowledge, has proposed pointed and operational information hiding perceptual model.6 Perceptual ModelThe perceptual model carries on analysis to information hiding object, describes information hiding basic question using through human vision perceptual characteristic and matrix theory. The perceptual model carries on description to information hiding process, thus it has established information hiding appraisal quantitative analysis. Distinguishes to other theoretical models is, incompletely borrows communication and information theory conclusion, the model has pointed and may be operational.6.1 Human perceiveThe human eye is extremely sensitive in wavelength, which range from 400 to 770 electro -magnetic radiations. The color image may use function to express. This function can express position, in reflected light wave length and the dynamic image situation time. Regarding color vision, the spectrum sensitivity function sum which some three foundations may use. The object carries on perceptual through human eye or photographic camera system to the scene in mainly passes its radiation. Between illumination and subjective human perceptual, as well as responds with humanity has direct relation. In the Weber law, first formulation has produced this kind of relations. The formula expression is:The experiment investigates indicated that Weber decides law to be only suitable for middle illumination value, low illumination value is not suitable very much regarding highest and lowest illumination. But perceptual brightness and with illumination relations is logarithm relations. According to Thomas Young tertiary color theory, our vision system perceptual all colors are basic color linear combinations. If between two kind of colors exists slightly may just noticeable difference, actually smallest may sense difference value not be constant, this mainly because of human vision non-linearity and the RGB space uniformity. If person eye color perceptual ability also not saturated, when stronger illumination, human eye color perceptual ability is better. Buchsbaum has considered eye adaption condition and illumination condition in of non-linear equality constant. Buchsbaum started his research in Weber research’s foundation analysis has obtained logarithm form visual non-linearity.6.2 Estimation Basic FactorsIt use a function to realize theperceptual model, after this function is calculates, the source image and information hiding between image perceptual distance, the traditional use MSE mean-square deviation is simplest distance function. The perceptual model basic factors are sensitivity, concealed and merge.(1)Sensitivity: Regarding visual perceptual, with input signal frequency correlation. The frequency response mainly refers to spatial frequency, spectral frequency and temporal frequency. The spatial frequency usually refers to brightness sensitivity, spectral frequency by color form perceptual, generally is low-frequency response, temporal frequency response by movement form perceptual.(2) Concealed: there has two kinds in vision concealed, frequency concealed and brightness concealed, the former is section of frequencies, another section of frequencies perceptual concealed, the latter is partial brightness, which meets the2-230concealed contrast gradient change.(3)Merges: it is away from in perceptual model, turns over to many different distorted perceptual synthesis one for to covert carrier appraisal, is called merge, the formula is:(4) merges, compares to the primitive image and the covert image, the computation corresponds the DCT coefficient the bad value, then team by contrast gradient concealed, obtains each item to be possible the perceptual distance, then merges an always perceptual distance is:6.3 Watson Perceptual ModelWatson proposed a survey vision fidelity model, it had estimated between source image and covert image JND value, joins influence after noise which image produces to estimate, this model must be better than MSE effect. The model basic principle is separated DCT estimate change according to image block perceptual, then double perceptual distance individually estimate these estimate merges. This model based on DCT domain, most melts processing after piecemeal, image energy concentrates in each together low frequency part, like this can estimate quantification noise perceptual, may act according to each image concrete characteristic to change equation length of stride. The present paper needs to use this model appraisal and control information hiding algorithm The Watson model by an sensitivity function, two concealed and merge partially is composed based on brightness and contrast gradient.(1)the model defined frequency sensitivity table, in the table each element expressed in each when did not have any concealed noise, but perceptual DCT coefficient smallest scope value.(2)Brightness concealed, if DCT block average brightness value higher, a DCT coefficient bigger revision is not easy to realize, the Watson model to each image element block, carries on the adjustment according to sensitivity table value, brightness concealed threshold valueis:TL C k C j i t k j i t D )/],0,0[](,[],,[0,00 (3)The above brightness concealed value also comes under contrast gradient concealed influence,the contrast gradient concealed threshold value is:6.4 Watson Perceptual Model ImprovementThis model exist many defects, which must rely on correct synchronization and overestimate translation effect, simultaneously also can underestimate influence which certain noises brings. Therefore in view of to Watson perceptual model, the present paper has pointed performs improvement. May to template which joins carry on according to perceptual model adjusts realizes auto-adapted Watson perceptual model. In other words acts according to anticipated perceptual to beaway from to embedment strength carries on automatic control. Proposed in Podilchuk and Zeng in algorithm foundation, we proposed one kind inserts based on perceptual forming Watson perceptual model. It acts according to perceptual distance which model surveys to carry on embed, simultaneously chooses embed different intensity, simultaneously the information symbolized information for changes into perceptual forming symbol. Improvement model as follows:Fig 6 Based on Perceptual Formed Perceptual Model2-2317 ConclusionThe present paper based on existing information hiding model analysis, had pointed out theoretical model existence shortcoming, in perceptual model foundation which proposed according to Watson, improved theoretical model, from fundamental research, has opened up the research information hiding model mentality, simultaneously the further work consummates this model, then carries on modeling ,and further quantitative analysis.Reference[1]IJ.Cox, Watermarking as communications with side information[j], procedding of the IEEE, 1999:1127-1141.[2] C Cachin, An imformation theoretic model for steganography[c], proceeding of the 2nd workshop on information hiding,lecture notes in computer science, springer,1998,57-78.[3]H. V. Poor, An introduction to signal detection and estimation, Springer, NY, 2000[4]P. Moulin and J.A. O’sullivan, “Information-theoretic analysis of Steganography,” preprint, Dec. 2001[5]K. Sullivan, O. Dabeer, U. Madhow, B. S. Manjunath, andS. Chandrasekaran, “LLRT based detection of LSB hiding,” submitted to ICIP 2003Author BiographyGe xiuhui˖ Female, engineer, Doctor of Engineering, research direction: Network security direction, information hiding.2-232。