Implementation of Image Steganography Using Ideal Segmentation Of Bit - Planes (Research Paper Sample)

Implementation of Image Steganography Using Ideal Segmentation Of Bit - Planes Aayushi Tiwari Computer Science and Engineering Vellore Institute of Technology Vellore, India aayushirenu17@gmail.com Abstract—Bit Plane Complexity Segmentation (BPCS) steganography is a technique used to hide secret data within an image by manipulating its bit plane complexity. The method works by dividing the image into non - overlapping segments and analyzing the complexity of each segment in terms of the number of 1’s and 0’s in its bit plane representation. The more complex a segment, the more information it can potentially hide. One of the limitations of the original BPCS method is that it can be detected by statistical analysis of the image. Researchers have proposed several improved methods to address this issue, including the use of adaptive segmentation, multiple embedding levels, and enhanced compression techniques. This paper proposes an improved method - Adaptive Multi-Level BPCS (AML-BPCS) technique, which uses multiple bit planes and multiple embedding levels to achieve a higher level of security. In this technique, the complexity of each segment is analyzed at different bit planes, and the data is embedded in the bit plane with the highest complexity. This makes it more difficult for attackers to detect the presence of hidden data. Index Terms—Adaptive; BPCS; DCT; Complexity; Bit Planes * INTRODUCTION Steganography is the means by which a message is repre- sented within an object, such as an image, text file, audio and video files, in such a way that the presence of the existence of the message is not known under normal human inspection. The message is hidden in such a way that only the receiver, knowing the algorithm, can decode it. The method allows one party to communicate with another party without anybody else being aware about any exchange of information happening. The advantage of steganography is that the transfer object does not arouse any suspicion when viewed normally, and the secret message cannot be extracted without knowing the encoding- decoding algorithm. In steganography, the secret message that is hidden must possess features such as good hiding capacity, perceptual transparency, robustness and tamper resistant. Old techniques such as Least Significant Bit (LSB) Technique, have the hiding capacity of only 10-15% of the secret data. In LSB technique, secret data is hidden in the last four least significant bits, due to which the overall quality of the vessel image reduces, which makes it vulnerable to the hackers. In order to overcome this limitation, BPCS Steganography Technique was introduced. * BIT PLANE COMPLEXITY SEGMENTATION STEGANOGRAPHY * Basic Principle The basic principle of BPCS Steganography is that the vessel image is divided into “informative” and “noise-like” regions, where one of them is simple whereas the other is complex. The secret data is also divided into informative and noise-like regions. In the hiding process, the noise-like region of vessel image is interchanged with that of secret data, due to which the image quality of the vessel image is not altered at all and nobody can suspect that a secret data is hidden inside it, therefore, implementing this nature of human vision system. The vessel image, consisting of p pixels is split into n binary images, also called Bit Plane Slicing. For eg. - N is an image with p=8 pixels. Therefore P is split into [N7 N6 N5 N4 N3 N2 N1 N0], where N7 being the most significant bit and N0, the least significant bit. In order to keep the picture quality of the vessel image intact, the secret data is hidden in the noise-like regions of both the most significant as well as least significant bit. Before splitting the vessel image, it’s pixels, which are originally in Pure Binary Code (PBC), is converted into Canonical Gray Code (CGC). * The Theory of Canonical Gray Coding in BPCS Gray scale images are black and white images and are composed of 0’s and 1’s, where 0 being black, with the least intensity and 1 being white, with the most intensity. Ex-oring of bits is done in this technique. PBC is converted to CGC for hiding the secret data. The reason is that PBC suffers from “Hamming Cliff”, meaning that a small change in colour affects many bits of colour value. For eg. - there are two consecutive pixels having values 127 and 128. In PBC, 127 is represented as 01111111 and 128 is represented as 10000000. When secret data is embedded, then 01111111 can change to 11111111 and 10000000 can change to 00000000. Due to this change, one pixel will appear pure black whereas the other pixel will appear pure white after embedding, which will easily be visible to the naked eyes. CGC doesn’t suffer from Hamming Cliff, due to it’s technique of ex-oring the bits. * Bit Plane Slicing in BPCS The process of dividing the vessel image into it’s elemental binary planes is called Bit Plane Slicing. Bit Plane Slicing is done in order to find out the noise-like regions for embedding of secret data into the vessel image. The vessel image is split into binary planes in the multiples of 12, and subsequently, the bit-planes are split into 8x8 blocks. * Complexity Measure For Embedding Secret Data After Bit Plane Slicing is done, the noise-like regions, also known as complex regions is to be found out. This process is called Complexity Measure. There are many methods of complexity measure in BPCS Steganography, the most used and easy one being black and white border length method. The formula to calculate complexity through this method is - α = k/(2 ∗ 2m ∗ (2m − 1)(1) where k denotes total length of border in the image. * Conjugation Operation Of Secret Data After complexity is calculated, Conjugation Operation is performed on the secret data in order to hide it in the vessel image. Since the secret data has to be embedded in the noise- like regions of the vessel image, the Conjugation Operation is performed in order to transform the informative patterns of the secret data into noise-like patterns. The already existing noise-like patterns are directly replaced with that of vessel image without any operation performed. Conjugate is calculated using the formula - α* = 1 − α(2) * PROPOSED METHODOLOGY The existing BPCS Steganography technique has some flaws associated with it . The method of calculating complexity is very easy and crackable, and the hackers can retrieve the secret data. Also, the secret data can be detected using statistical analysis of the vessel image. The BPCS Steganography is dependant on the Complexity Measure, and there should be a strong algorithm devised for calculation of complexity. * Adaptive Multi-Level BPCS Steganography Using Discrete Cosine Transform This paper puts forward an improved BPCS Steganography, which proposes a take on the method of Complexity Measure and embedding of secret data. In the proposed method, the Complexity Measure will be calculated using Discrete Cosine Transform (DCT) and the Conjugation Operation will be per- formed subsequently using Inverse Discrete Cosine Transform. For the embedding of secret data, the bit planes will be sorted according to their highest complexity, and the secret data will be hidden in the LSB of the highest complexity. This will ensure that the secret data is embedded in the deeper level of the vessel image and won’t be caught during steganalysis. * Discrete Cosine Transform In Image Processing Discrete Cosine Transform (DCT) is used to compress images for embedding secret data. It separates the image into parts of differing importance. It can separate the image into high, middle and low frequency components. The equation for a 1D DCT is - —−C(u) = α(u)Σ(x = 0> N1)f (x)cos[(2x + 1)uπ/2N ] (3) for u = 0,1,2,.,N-1. DCT is applied to each block of each bit plane obtained after Bit Plane Slicing of the vessel image. Each block is then compressed through quantization table to scale the DCT coefficients and the secret data is hidden in the DCT coefficients. * Adaptive Multi-Level Theory Adaptive Multi-Level is a proposed method in this paper in BPCS Steganography for hiding of secret data. In this method, the bit-planes of the vessel image so obtained after Bit-Plane Slicing, also called levels of embedding, is counted, and for each level of embedding, the bit-plane containing the highest complexity is selected, and the secret data is hidden in the LSB of the selected bit-planes. The reason why this method is proposed is because anything hidden in the higher complexity regions is not visible and detected easily by any steganalysis technique, and since LSB is the lowermost bit and there are 6 more levels above the LSB, the secret data is actually hidden in the deeper levels of the higher complexity region, hence dismissing the doubts of any secret data being a part of the vessel image. * Algorithm Proposed We would be using an RGB image as the vessel image for embedding secret data. The secret data will also be an RGB image for ease of embeddment. An RGB image constitutes of three different images - red, green and blue images, combined one after the other respectively. The images, when viewed together, are seen as a colour image. The RGB image is made up of 3 mxn matrices, each of red, green and blue, and the range od each colour ranges from 0-255, where 0 is the lowest and 255 is the highest. A colour image is the combination of the different ranges of each of the three colours. Do Bit-Plane Slicing by dividing both the vessel image and the secret data into bit-planes and each bit plane into 8x8 blocks. Each of the three matrices of the RGB image should be divided into 24 bit-planes respectively. The number of bit- planes and blocks can be increased as desired. Apply DCT to each block and quantize the DCT coefficients using a quantization matrix. The quantization matrix is a fixed matrix th...