Steganalysis for Calibrated and Lower Embedded Uncalibrated Images

The general purpose of steganalysis algorithms is to detect a secret message in digital images. Nevertheless, previous studies have rather ignored the fact that more valuable information like message length can be extracted from stego images. In this article, a method is proposed for estimating message length in stego images resulted from PQ, which is a common steganography algorithm. In this method, the stego image resulted from PQ algorithm is first calibrated so that the corresponding cover image is obtained. Then, threshold μ is compared to the changes of global histograms of cover and stego images. If the threshold is less than the changes, first-order statistics and global histogram will be used to estimate message length; otherwise, secret message length will be estimated using second-order statistics and changes of variation of cover and stego images.

Steganography is a process to conceal the existence of a message. Image Steganography involves hiding the message onto a cover image to obtain a resultant stego image. This paper discusses a procedure for hiding coloured message images onto coloured cover images using a lossy Image Steganography technique. The secret images undergo Discrete Cosine Transformation, quantization and AES encryption before embedding using the Least Significant Bit embedding (LSB) technique onto the cover image. Finally, Peak signal-to-noise ratio (PSNR), Structural Similarity Index measurement (SSIM), zero normalized cross-correlation (ZNCC) are used for comparing the cover and stego images as well as the extracted and original images.

Steganography, the technology of protecting a secret message by embedding it inside a cover image, continues to be investigated and enhanced as an alternative data protection method. This paper deals with hiding multimedia files in true color RGB cover images with an emphasis on reducing the cover size, increasing hiding capacity and enhancing security of the hidden data. A proposed model (DuoHide) is presented in which a secret multimedia file, regardless of its type, is processed without un-compression, and divided between two cover images of equal size and dimensions. The multimedia file is read as a stream of bytes and split vertically into two parts, one part contains the least significant half-bytes, and the other part contains the most significant half-bytes. The two parts are hidden inside two uncompressed RGB cover images using a least significant 4-bit replacement technique. The resulting dual stego images are expected to be sent separately, through different channels, to avoid capture of both stego files by an adversary. Extraction of the secret file is achieved through merging LSB half-bytes from the two stego files. The extracted file is identical in content and structure with the original secret file. The implemented DuoHide system was evaluated using a set of public multimedia files, images, audios, and videos, of various sizes. The secret file sizes ranged from 5% to about 100% of the cover image's size. The experimental results showed that even at the highest embedding ratio, which is based on the secret-to-cover ratio, there were no perceptible visual differences between cover and stego images. The PNSR value was calculated as PSNR1, for cover1 and stego1, and PSNR2 for cover2 and stego2. The lowest PSNR value was around 31 dB for the highest embedding ratio, which is considered acceptable concerning statistical imperceptibility. The PSNR value increased as the embedding ratio decreased, reaching around 65 Decibel (dB) for the case of 5% secret-to-cover ratio. The integrity of the extracted secret file was verified through a bitwise comparison between original and extracted files, which showed zero differences. The DuoHide model is expected to provide better security for the hidden file, in case an attacker manages to capture one of the stego images and recover the hidden content because the attacker will only get an incomprehensible set of half-byte bits. An additional advantage of using a pair of stego files is that of reducing stego file size by 50%, to avoid problems and limitations of transmitting large files, especially that multimedia files are often large, and they cannot be compressed because they are already compressed. Security of the DuoHide system can further be improved by randomizing storage locations within the two stego images.

Steganography deals with the ways of hiding communicated data in such a way that it remains confidential. Finding best position inside cover image to embed text message, maintaining a reasonable trade-off between security, robustness, higher bit embedding rate and imperceptibility are some of the challenges of steganography system. Hence, this paper presents firefly algorithm for finding best positions inside cover image in order to embed text message into cover image using Pixel Value Differencing (PVD) technique. Four different cover image was used. Experimental result showed the cover image with selected location using firefly algorithm as well as the stego image using PVD technique. The stego image was evaluated using Peak Signal to Noise Ratio (PSNR) and Mean square Error (MSE). Firefly Algorithm with PVD technique produced a promising result for image steganography.

Data is an asset for all parties. So that it must always be guarded or secured so that it is not misused by irresponsible parties. One of the methods used for data security is steganography. In this paper, efforts to secure data are discussed through the development of steganographic algorithms using the Haar integer wavelet transform (IWT-Haar) and least significant bit (LSB) techniques to hide secret messages in the form of digital text into digital images (cover images) to produce stego images. The results show that the extracting process is much faster than the embedding process. The PSNR values obtained ranged from 55-71,5 dB (far greater than 40 dB). This shows that the resulting stego image is of very good quality. In plain view, the difference between the cover image and the stego image has not changed significantly.Key words: digital image, embedding, extraction, Haar integer wavelet transform, steganographyÂ

R261, Jana et al. proposed the partial reversible data hiding (PRDH) based on (7, 4) Hamming code, which deals with three images: the original image (OI), the cover image (CI), and the stego image (SI). The CI is obtained by slightly modifying OI. After that, one can embed secret into CI to generate SI. The so-called partial reversible feature is that we can reconstruct the CI faultlessly, which is almost the same to the OI. However, in Jana et al.'s PRDH, the authors only adjust redundant bits by using (7, 4) Hamming code with even (or odd) parity. This simple approach may modify 3 redundant bits at most in 7 least significant bit (LSBs) to obtain CI from OI, and this degrades the CI quality seriously. In this paper, we construct two schemes: the proposed PRDH and the modified PRDH. The proposed PRDH enhances the partial reversible property that improves the visual quality of CI. And, the modified PRDH enhances the proposed PRDH to achieve the high embedding capacity. In addition, we also point out two inaccurate descriptions in Jana et al.'s PRDH about position of embedded secret to more simplify the procedure of embedding secret.

This paper presents a new secret sharing scheme for identification of cheaters in the elections. In a proposed scheme secret image (voter ID) is embedded into cover image (Indian emblem) of same size and generates n stego images. The resulting stego images are visually good and look like a cover image. Various image quality metrics are used to assess the quality of stego images. Out of n stego images only last two stego images are required to reconstruct the secret and cover images without any loss. The proposed scheme offers high embedding capacity. Experimental results of proposed scheme shown to be superior in terms of PSNR and visual quality than existing schemes.

This paper presents a novel reversible data hiding method, which can recover the original image content without distortion from the stego images after the secret data have been extracted. The method utilizes the original cover image to finally generate two visually similar stego images by embedding additional data. During the data embedding process, the pixels in the first stego image are utilized to embed secret data by applying the proposed prefix matching technique, which exploits each of the gray values to carry one or two bits only with ± 1 operation. Pixels in the second stego image are adaptively modified to carry rest secret data by referring to the first stego image using the directed LSB embedding. For the recipient, the additional data can be extracted from the two stego images with the DH key, and the original cover image can be recovered with the location map. If the cover image has no boundary gray-values, one can always recover the cover image from the two stego images without the aid of any extra information. Experiments show that the proposed method can achieve desirable embedding capacity and maintain a high level of image quality.

A new and universal steganalysis scheme for JPEG images was proposed. Moments in frequency domain of wavelet sub-bandpsilas histogram was extracted to analyze the difference of histogram between cover image and stego image, at the same time, used high-pass filter to strengthen the image in order to make the discontinuity among blocks become more obvious, so the extracted block characteristic will be more sensitive to the steganographic action. Then Support Vector Machine (SVM) was used to classify the extracted 40 dimensional feature vectors. The experimental results of 3 popular JPEG steganographic schemes (F5, Outguess and MB1) demonstrate that the proposed scheme has the advantage in detection rate and speed.

Many JPEG steganography techniques are used to communicate secret messages by terrorists that threaten the security of a nation. So steganalysis is very important. Avcibas proposed a steganalysis based on binary similarity measures which only work well on LSB-based steganography and derived features from the spatial domain of images. This paper proposes a novel usage of binary similarity measures in JPEG steganalysis. The method captures the seventh and eighth bit planes of the non-zero DCT coefficients from JPEG images and computes 14 features of each image based on binary similarity measures. These features are used to construct a support vector machine classifier which can distinguish between stego images and cover images. The experiment results are presented to demonstrate that the proposed scheme has lower computational complexity and the same high detecting accuracy.

This paper proposes an active steganalysis method for quantization index modulation (QIM)-based steganography. The proposed nonparametric steganalysis method uses irregularity (or randomness) in the test image to distinguish between the cover image and the stego image. We have shown that plain quantization (quantization without message embedding) induces regularity in the resulting quantized object, whereas message embedding using QIM increases irregularity in the resulting QIM-stego. Approximate entropy, an algorithmic entropy measure, is used to quantify irregularity in the test image. The QIM-stego image is then analyzed to estimate secret message length. To this end, the QIM codebook is estimated from the QIM-stego image using first-order statistics of the image coefficients in the embedding domain. The estimated codebook is then used to estimate secret message. Simulation results show that the proposed scheme can successfully estimate the hidden message from the QIM-stego with very low decoding error probability. For a given cover object the decoding error probability depends on embedding rate and decreases monotonically, approaching zero as the embedding rate approaches one.

Feature Selection is a preprocessing technique with great significance in data mining applications that aims at reducing computational complexity and increase predictive capability of a learning system. This paper presents a new hybrid feature selection algorithm based on Discrete Firefly optimization technique with dynamic alpha and gamma parameters and t-test filter technique to improve detectability of hidden message for Blind Image Steganalysis. The experiments are conducted on important dataset of feature vectors extracted from frequency domain, Discrete Cosine Transformation and Discrete Wavelet Transformation domain of cover and stego images. The results from popular JPEG steganography algorithms nsF5, Outguess, PQ and JP Hide and Seek show that proposed method is able to identify sensitive features and reduce the feature set by 67% in DCT domain and 37% in DWT domain. The experiment analysis shows that these algorithms are most sensitive to Markov features from DCT domain and variance statistical moment from DWT domain.

A novel algorithm for colour image steganography using a new intelligent technique based on three phases

With better steganographic methods, steganalysis is a skill of finding concealed information within digital media. It has grown even more important now a days. In this work, a deep learning-based method to identify stego images by means of ResNet50V2 convolutional neural network architecture is proposed. With equal representation of cover and stego images, the dataset consisted in 6,602 training and 665 testing images scaled to 150×150 pixels. We used data augmentation to improve model resilience. With a precision of 65.42%, recall of 91.76%, and Fl-score of 0.76 the model attained an overall accuracy of 87.52%. Strong performance in identifying stego images (class 1) with a recall of 92% and Fl-score of 0.93 indicates by detailed classification metrics the usefulness of the model in separating hidden material. Cover image (class 0) detection showed reduced precision and specificity, nonetheless, which indicates space for development. These findings open the path for more optimization in forensic and cybersecurity uses and validate the potential of ResNet50V2 in steganalysis activities.

A reversible extended secret image sharing scheme based on Chinese remainder theorem