LITERATURE SURVEY: Steganography Using Redundant Bit Replacement By Neural Network Jasmeet Kaur 1, Nitika Kapoor 2, Harish Kundra 3

Transcription

1 LITERATURE SURVEY: Steganography Using Redundant Bit Replacement By Neural Network Jasmeet Kaur 1, Nitika Kapoor 2, Harish Kundra 3 1 Research Scholar, 2,3 Assistant Professor 1,2,3 Department of Computer Science and Engineering, Rayat Institute of Engineering and Information Technology Railmajra, SBS Nagar, (Punjab) INDIA 1 jasmeetropar@gmail.com, 2 er.nitikakapoor@gmail.com, 3 hodcseit@rayatbahra.com Abstract: - The paper describes the progress in the field of Steganography. The idea behind this technique is to hide the information in the media. The challenge is to make the hidden information untraceable. The concept originate from spatial domain to more enhanced technique. The proposed technique in this paper is Neural Network. By using this technique we make hide the information in better way than simpler techniques in spatial domain. I. INTRODUCTION Classic methods of securing communication mainly base on cryptography, which encrypts plain text to generate cipher text. However, the transmission of cipher text may easily arouse attackers suspicion, and the cipher text may thus be intercepted, attacked or decrypted violently. In order to make up for the shortcomings of cryptographic techniques, steganography has been developed as a new covert communication means in recent years. It transfers message secretly by embedding it into a cover medium with the use of information hiding techniques. Cryptography and Steganography are two important branches of information security. Cryptography provides encryption techniques for a secure communication. Cryptography is the science that studies the mathematical techniques for keeping message secure and free from attacks [6]. Steganography is the art and science of hiding communication. The word steganography is derived from the Greek word stegos meaning cover and grafia meaning writing defining it as covered writing.steganography involves hiding information so it appears that no information is hidden at all. Steganalysis is the science of detecting hidden information. The goal of steganalysis is to break steganography. Steganalysis deals with three important attacks. (a) Visual attacks: one can identify the stego image with the naked eyes (b) Statistical attacks: they reveal the smallest alterations in an image statistical behaviour. It is further subdivided into (i) Passive attack: identifying the presence or absence of a covert messages or embedding algorithm used (ii) Active attacks: used to investigate embedded message length or hidden message location or secret key used in hidden process (c) Structural attacks: identifying the changes in the cover file. Steganography is employed in various useful applications, e.g., copyright control of materials, enhancing robustness of image search engines and smart IDs (identity cards) where individuals details are embedded in their photographs. Other applications are video-audio synchronization, companies safe circulation of secret data, TV broadcasting, TCP/IP packets (for instance a unique ID can be embedded into an image to analyze the network traffic of particular users), and also checksum embedding [8]. One method of common Steganography technique is to hide the secret message in the least significant bits of pixels of the cover image. The image quality of stego image achieved by applying the LSB technique is very closer to the original one. But the drawback is it cannot survive image processing manipulations. One method of LSB Steganography involves manipulating the LSB plane from direct replacement of the cover image with message bits to some type of logical or arithmetic combination between two. Several examples of LSB techniques are found. This technique achieves both high capacity and low perceptibility. But it is not very sophisticated and subject to extraction by unwanted persons. Masking and filtering techniques usually restricted to 24 bits or grayscale images. These methods are effectively similar to paper watermarks, creating markings in an image. This can be achieved for example by modifying the luminance of parts of the image. While masking does change the visible properties of an image, it can be done in such a way that the human eye will not notice the anomalies. Least Significant Bit maintains a good visual quality of stego-image, it can hide little information. Considering the drawback of LSB, some methods begin to take account of the visual identity that human eyes are insensitive to edged and textured areas when embedding secret information, such as BPCS(biplane complexity segmentation),pvd(pixel value differencing), MBNS (multiple base notational system ), SOC, Side Match and WCL. The capacity of embedded information is thereby greatly improved while the quality of visual imperceptibility is maintained. As human vision sensitivity is complex, it is hard to exactly decide whether a pixel is in less sensitivity areas or not. Thus, based on the contrast and texture sensitivity, we train self-organizing map Neural Networks (NNs) trained to distinguish pixels in less sensitive areas from pixels in more sensitive areas. So, NNs trained is the secret key. Then, we use NNs trained to classify pixels, and select pixels in less sensitive areas to embed more secret data. On the receiving side, the original image is not needed for extracting the embedded data. Neural approach adds the complexity for the hackers accessing and also presents high potentiality in defense operations. Neural Steganography is a powerful tool that enables people to communicate without possible eavesdroppers even knowing there is a form of communication. Basic elements of steganography in images are shown in Figure 1. The carrier image in steganography is 35

2 called the "cover image" and the image which has the embedded data is called the "stego image". The embedding process is usually controlled using a secret key shared between the communicating parties. Fig:1 Typical Element Of Steganography System. II. RELATED WORK Adel Almohammad, Gheorghita Ghinea and Robert M. Hierons in 2009 proposed a High Capacity Steganographic Method Based Upon the JPEG standard which uses 8x8 quantization tables, but it does not specify any default or standard values for quantization tables. However, the JPEG standard provides a pair of quantization tables as examples tested empirically and found to generate good results. Dividing this quantization table they get a new quantization table. Using this new quantization table generates reconstructed images almost identical to the source image. Therefore, this table will be used with Jpeg-Jsteg method in the experiment. Since the values of these tables could be an arbitrary choice, some researchers modified these quantization tables for their research purposes. A quantization table can arbitrarily be generated. Consequently, They produced a 16x16 quantization table by simulating and stretching the scaled quantization table [3]. Kousik Dasgupta1, J.K. Mandal2 and Paramartha Dutta in April 2012 suggested a Hash Based LSB Technique for Video Steganography that deals with hiding secret data or information within a video. A spatial domain technique where the secret information is embedded in the LSB of the cover frames. Eight bits of the secret information is divided into 3,3,2 and embedded into the RGB pixel values of the cover frames respectively. A hash function is used to select the position of insertion in LSB bits. The proposed method is analyzed in terms of both Peak Signal to Noise Ratio (PSNR) compared to the original cover video as well as the Mean Square Error (MSE) measured between the original and steganography files averaged over all video frames. The proposed technique is compared with existing LSB based steganography and the results are found to be encouraging [1]. Some of few researchers have already implemented NEURAL NETWORK in their approach for the same here are some reviews about them. Usha B A1, Dr. N K Srinath2, Dr. N K Cauvery in May 2013 proposed a Data Embedding Technique using Neural Network. According to them the neural approach to embed information satisfies a secure steganography. Neural approach adds the complexity for the hackers accessing and also presents high potentiality in defense operations. Neural Steganography is a powerful tool that enables people to communicate without possible eavesdroppers even knowing there is a form of communication [5]. Imran Khan in August 2013 suggested an Efficient Neural Based Algorithm of Steganography for Image. To provide large capacity of the hidden secret data and to maintain a good visual quality of stego-image a novel steganography method based on neural network and random selection of edged areas of pixels is proposed. Firstly a cover image is divided into a non-overlapping two pixels block and this pixel block generates a set of edged non-overlapping regions. After this a neural network is applied which generates a stego-image which has been immune against conventional attack and performs good perceptibility compared to other steganography approaches. From our experimental results it can be shown that the proposed method hides information in edged regions and maintains a better visual display of steganography image than the traditional methods [8]. Bhavneet Kaur, Pooja & Harish Kundra in December2013 proposed a Performance Enhancement of a Transform Domain based Steganograhic Technique using Segmentation. This Method involves combining the DCT algorithm along with NEURAL NETWORK in such a way that the IMAGE QUALITY which is measured in terms of PSNR increases and the data remains safe within the image [9]. III. TAXONOMYOF STEGANOGRAPHIC TECHNIQUES There are quite a lot of approaches in classifying steganographic techniques. These approaches can be classified in accordance with the type of covers used with secret communications [10]. Steganographic techniques that modify image files for hiding information include the following: Spatial domain Transform domain Distortion techniques File Embedding technique 1. SPATIAL DOMAIN TECHNIQUE Spatial domain steganography techniques, also known as substitution techniques, are a group of relatively simple techniques that create a covert channel in the parts of the cover image in which changes are likely to be a bit scant when compared to the human visual system (HVS).One of the ways to do so is to hide information in the least significant bit (LSB) of the image data. 1.1LSB TECHNIQUE: This embedding method is basically based on the fact that the least significant bits in an image can be thought of as random noise, and consequently they become not responsive to any changes on the image. The disadvantage of this technique is that it uses each pixel in the image. As a result, 36

3 if lossy compression is used, some of the hidden information might be lost [10]. Limitation of LSB LSB technique in the spatial domain is a practical way to conceal information but, at the same time, it is vulnerable to small changes resulting from image processing or lossy compression [7]. Although LSB techniques can hide large quantities of information i.e., high payload capacity, they often compensate the statistical properties of the image and thus indicate a low robustness against statistical attacks as well as image manipulation. 2. TRANSFORM DOMAIN TECHNIQUE Transform domain embedding can be defined as a domain of embedding techniques for which a number of algorithms have been suggested. The process of embedding data in the frequency domain of a signal is much stronger than embedding principles that operate in the time domain. It is worth saying that most of the strong steganographic systems today operate within the transform domain. Transform domain techniques have an advantage over LSB techniques because they hide information in areas of the image that are less exposed to compression, cropping, and image processing. Some transform domain techniques do not seem dependent on the image format and they may outrun lossless and lossy format conversions. 2.1 JPEG COMPRESSION: If an image is to compress into JPEG format, the RGB color space is first turned into a YUV representation. Through this representation, the Y component represents brightness (or luminance) and the U and V components stand for color (or chrominance). It is known that the human eye is more sensitive to changes in the brightness of a pixel than to changes in its color. Down sampling the color information is taken as an advantage of the JPEG to reduce the size of the file where the color components (U and V) are splitted in the horizontal and vertical directions and consequently reducing the file size by a factor of 2. Then, the image is transformed. For JPEG images, the discrete cosine transform (DCT) is used; the pixels can be converted with such mathematical processing by simply spreading the position of the pixel values over the image or part of it [12]. With DCT transformation, a signal is transformed from the representation of an image into the frequency domain, this is done by sorting the pixels into (8 8) pixel blocks and transforming these blocks into 64-DCT coefficients which are affected by any modification of a single DCT coefficient. 2.2 Wavelet transform technique: Wavelets transform (WT) converts spatial domain information to the frequency domain information. Wavelets are used in the image steganographic model because the wavelet transform clearly partitions the highfrequency and low-frequency information on a pixel by pixel basis. The discrete wavelet transform (DWT) method is favored over the discrete cosine transform (DCT) method, owing to the resolution that the WT provides to the image at various levels.wavelets are mathematical functions that divide data into frequency components, which makes them ideal for image compression. In contrast with the JPEG format, they are far better at approximating data with sharp discontinuities. Researchers use vector quantization, called Linde-Buzo-Gray (LBG), associated with block codes, known as BCH codes, and one-stage discrete Haar wavelet transforms. They emphasize that modifying data by using a wavelet transformation produces good quality with few perceptual artifacts. A group of scientists at Iowa State University are developing an advanced application called artificial neural network technology for steganography (ANNTS), with the aim of detecting all current steganography methods, which include DCT, DWT, and DFT. They found that the inverse discrete Fourier transform (IDFT) includes a rounding error that makes DFT inappropriate for steganography applications [12]. The promising techniques such as DCT, DWT and the adaptive steganography are not tended to attacks, especially when the hidden message is small. This can be justified in relation to the way they change the coefficients in the transform domain, thus, image distortion is kept to a minimum. Generally speaking, such techniques tend to have a lower payload when they are compared to the spatial domain algorithms [8]. The experiments on the discrete cosine transform (DCT) coefficients have introduced some promising results and then they have diverted the researchers attention towards JPEG images. Working at some level like that of DCT turns steganography much more powerful and less prone to statistical attacks. Embedding in the DWT domain reveals a sort of constructive results and outperforms DCT embedding, especially in terms of compression survival 3. DISTORTION TECHNIQUES Distortion techniques require knowledge of the original cover image during the decoding process where the decoder functions to check for differences between the original cover image and the distorted cover image in order to restore the secret message. The encoder, on the other hand, adds a sequence of changes to the cover image. So, information is described as being stored by signal distortion. Using this technique, a stego-object is created by applying a sequence of modifications to the cover image. This sequence of modifications is selected to match the secret message required to transmit. The message is encoded at pseudo-randomly chosen pixels. If the stego-image is different from the cover image at the given message pixel, then the message bit is a 1. [10] Otherwise, the message bit is a 0. The encoder can modify the 1 value pixels in such manner that the statistical properties of the image are not affected (which is different from many LSB methods). However, the need for sending the cover image limits the benefits of this technique. 4. FILE EMBEDDING TECHNIQUE Different image file formats are known for having different header file structures. In addition to the data values, such as pixels, palette, and DCT coefficients, secret information can also be hidden in either a header structure or at the end of the file [12]. For example, the comment fields in the header of JPEG images usually contain data hidden by the invisible Secrets and 37

4 aa s ISSN Steganozorus. Camouflage, JpegX, PGE10, and PGE20 add data to the end of a JPEG image. Limitation Of File Embedding Technique File formatting techniques can store large amounts of information, but they are easily detected and attacked. LSB TRANSFORM DOMAIN FILE EMBEDDING IMPERCEPTIBILITY High High High Low ROBUSTNESS Low High Low Low PAYLOAD CAPACITY High Low High Low DISTORTION Table 1: Comparison Of Different Techniques. IV. TECHNIQUE INVOLVED NEURAL NETWORK An Artificial Neural Network (ANN) is an information processing paradigm that is inspired by the way biological nervous systems, such as the brain, process information. The key element of this paradigm is the novel structure of the information processing system. It is composed of a large number of highly interconnected processing elements (neurons) working in unison to solve specific problems. ANNs, like people, learn by example. An ANN is configured for a specific application, such as pattern recognition or data classification, through a learning process. Learning in biological systems involves adjustments to the synaptic connections that exist between the neurons. This is true of ANNs as well. Why use neural networks? Neural networks, with their remarkable ability to derive meaning from complicated or imprecise data, can be used to extract patterns and detect trends [8] that are too complex to be noticed by either humans or other computer techniques. A trained neural network can be thought of as an "expert" in the category of information it has been given to analyze. This expert can then be used to provide projections given new situations of interest and answer "what if" questions. Other advantages include: 1. Adaptive learning: An ability to learn how to do tasks based on the data given for training or initial experience. 2. Self-Organization: An ANN can create its own organization or representation of the information it receives during learning time. 3. Real Time Operation: ANN computations may be carried out in parallel, and special hardware devices are being designed and manufactured which take advantage of this capability. 4. Fault Tolerance via Redundant Information Coding: Partial destruction of a network leads to the corresponding degradation of performance. However, some network capabilities may be retained even with major network damage. A simple neuron An artificial neuron is a device with many inputs and one output. The neuron has two modes of operation; the training mode and the using mode. In the training mode, the neuron can be trained to fire (or not), for particular input patterns. In the using mode, when a taught input pattern is detected at the input, its associated output becomes the current output. If the input pattern does not belong in the taught list of input patterns, the firing rule is used to determine whether to fire or not. A simple neuron V. CONCLUSION AND FUTURE WORK This paper provides an overview of steganography & reviewed the main steganographic techniques. Each of these techniques tries to satisfy the three most important factors of steganographic design (imperceptibility or indefectibility, capacity, and robustness). We can deduce that while one technique may lack in payload capacity, another may lack in robustness. For example, file formatting techniques can store large amounts of information, but they are easily detected and attacked. Likewise, LSB techniques in a spatial domain have a high payload capacity, but they often fail to prevent statistical attacks and are thus easily detected. Besides, file and spatial domain approaches are considered not to be robust against lossy compression and filtering. Transform domain techniques are considered more robust for lossy compression image formats, but this advantage is achieved at the expense of payload capacity. However, it is possible to defeat the transform domain techniques using Neural Network. Neural network generates a stego-image which has been immune against conventional attack and performs good perceptibility compared to other steganographic approaches. 38

5 REFERENCES [1] Kousik Dasgupta1, J.K. Mandal2 and Paramartha Dutta",HASH BASED LEAST SIGNIFICANT BIT TECHNIQUEFORVIDEOSTEGANOGRAPHY(HLSB)Int ernational Journal of Security, Privacy and Trust Management IJSPTM), Vol. 1, No 2, April 2012 [2] Shamim Ahmed Laskar1 and Kattamanchi Hemachandran2 High Capacity data hiding using LSB Steganography and Encryption International Journal of Database Management Systems ( IJDMS ) Vol.4, No.6, December 2012 [3] "Adel Almohammad Robert M. Hierons" High Capacity Steganography Method Based Upon JPEG The Third International Conference on Availability, Reliability and Security The JPEG standard uses 8x8 quantization tables, [4] "Ross J. Anderson, Fabien A.P. Petitcolas" On The Limits of Steganography IEEE Journal of Selected Areas in Communications, 16(4): , May [5] "Usha B A1, Dr. N K Srinath2, Dr. N K Cauvery" DATA EMBEDDING TECHNIQUE INIMAGE STEGANOGRAPHY USING NEURAL NETWORK International Journal of Advanced Research in Computer and Communication Engineering Vol. 2, Issue 5, May 2013 [6] " Ms. P. T. Anitha1, Dr. M. Rajaram2,Dr. S. N. Sivanandham" AN EFFICIENT NEURAL NETWORK BASED ALGORITHM FOR DETECTING STEGANOGRAPHY CONTENT IN CORPORATE MAILS: A WEB BASED STEGANALYSIS IJCSI International Journal of Computer Science Issues, Vol. 9, Issue 3, No 1, May 2012 [7] "Nameer N. EL-Emam " Efficient Steganography using NEURAL [8] "Imran Khan" International Journal of Computer Technology and Electronics Engineering (IJCTEE) Volume 1, Issue 2 [9] Bhavneet Kaur, Pooja & Harish Kundra PERFORMANCE ENHANCEMENT OF A TRANSFORM DOMAIN BASED STEGANOGRAPHIC TECHNIQUE USING SEGMENTATION International Journal of Advances in Science and Technology Vol. I, Issue I, December 2013 [10]Nagham Hamid, University Malaysia Perils (UniMAP) nagham_fawa@yahoo.com School of Communication and Computer Engineering Penang, Malaysia. Topic: Image Steganography Techniques: An Overview. [11] Atallah M. Al-Shatnawi, Vol. 6, 2012, no. 79, Department of Information Systems Alalbayt University, Mafraq, Jordan. Topic: A New Method in Image Steganography with Improved Image Quality. [12]L.D. Paulson. (2006, Aug.). New system fights steganography. News briefs. IEEE Computer Society. [On line].39(8),pp available: [Jul., 2011]. 39