Non-Local Graph-Based Prediction for Reversible Data Hiding in Images

With the advent of 3D video compression and Internet technology, 3D videos have been deployed worldwide. Data hiding is a part of watermarking technologies and has many capabilities. In this paper, we use 3D video as a cover medium for secret communication using a reversible data hiding (RDH) technology. RDH is advantageous, because the cover image can be completely recovered after extraction of the hidden data. Recently, Chung et al. introduced RDH for depth map using prediction-error expansion (PEE) and rhombus prediction for marking of 3D videos. The performance of Chung et al.’s method is efficient, but they did not find the way for developing pixel resources to maximize data capacity. In this paper, we will improve the performance of embedding capacity using PEE, inter-component prediction, and allowable pixel ranges. Inter-component prediction utilizes a strong correlation between the texture image and the depth map in MVD. Moreover, our proposed scheme provides an ability to control the quality of depth map by a simple formula. Experimental results demonstrate that the proposed method is more efficient than the existing RDH methods in terms of capacity.

Reversible data hiding (RDH) is an active research area in the field of information security. The RDH scheme allows the transmission of a secret message by embedding it into a cover image, and the receiver can recover the original cover image along with the extraction of the secret message. In this paper, we propose a bit plane compression based RDH scheme to hide a sequence of secret message bits into a grayscale image. In the proposed method, a selected bit plane of the cover image will be compressed using run-length encoding (RLE) scheme. Further, the RLE sequence has been efficiently encoded as a binary sequence using Elias gamma encoding method. The Elias gamma encoded bit sequence concatenated with the secret message bits are used to replace the selected bit plane after performing a sequence of Arnold transform. The Arnold transform helps to find a new scrambled version of the bit plane which is very close to the original bit plane to ensure the visual quality of the stego image. The RLE is a lossless compression technique, therefore recovery of the original image is possible by the receiver. The experimental study of the proposed scheme on the images from standard image dataset (USC-SIPI image dataset) shows that the proposed scheme outperforms the existing scheme in terms of the visual quality of the stego image without compromising the data embedding rate.

Reversibility improved data hiding in 3D mesh models using prediction-error expansion and sorting

State-of-the-art data analysis tools have to deal with high-dimensional data. Fortunately, the inherent dimensionality of data is often much smaller, as it has an internal structure limiting its degrees of freedom. In most cases, this structure can be approximated using a graph, i.e, a set of nodes connected by edges. Based on this idea, graphs have been largely used in semi-supervised and unsupervised learning. The canonical assumption when incorporating a graph prior is to assume that the signal is smooth with respect to the graph, i.e, that connected pieces of data have similar values. Nevertheless, this hypothesis is insufficient to characterize more complex relations between the data and its structure, the graph. To this end, the field of Graph Signal Processing (GSP) extends the graph smoothness hypothesis to a more general assumption. Its key idea is to think of the signal as a sum of harmonic modes, which are obtained from the eigen-decomposition of the Laplacian, an operator representing the second order derivative. Thus, the essence of GSP resides in the Fourier transform defined by the projection on the Laplacian eigenvectors basis, which allows us to filter graph signals. However, GSP suffers from a counterintuitive irregularity. Contrarily to the classical Fourier basis, the graph Laplacian eigenvectors are not spreading uniformly on the vertex set and can be highly concentrated in a region of the graph. This non-uniformity results in a lack of any intuitive translation operator and consequently poses challenges to generalize some of the classical signal processing theory, e.g., stationarity, uncertainty principles, or sampling. In this thesis, we answer these difficulties by accepting that each node is unique and should be treated specially according to its surrounding structure. This is achieved using an operator called localization that shifts a kernel defined in the spectral domain around a specific node while adapting to the local graph structure. In the first part of the thesis, we focus on harmonic analysis in GSP. We start by making use of the norm of localized kernels to capture the spectral local features of the graph vertices. To illustrate the relevance of this analysis we then introduce structural clustering, an algorithm that groups nodes according to the role they play in the graph. Then, we bound the vertex-frequency concentration of graph filter banks atoms. Because of the irregular structure of graphs, we construct a local uncertainty principle that is, again, driven by the norm of a localized kernel. In the second part of the thesis, we tackle GSP problems from a machine learning perspective. We use the localization operator to extend the notion of stationarity to graph signals. It consists of a graph signal probabilistic framework allowing us to optimally solve inverse problems. Finally, we show how the graph total variation can be used in semi-supervised learning and prove its consistency with respect to the underlying graph manifold.

Reversible data hiding (RDH) is a technique that slightly alters digital media (e.g. images or videos) to embed secret messages while the original digital media can be completely recovered without any error after the hidden messages have been extracted. In the past more than one decade, hundreds of RDH algorithms have been reported, and among these algorithms, the difference histogram shifting (DHS) based methods have attracted much attention. With DHS-based RDH, high capacity and low distortion can be achieved efficiently. But there occurs one problem that, with DHS, the difference values to embed secret bits are explored, and the other difference values are shifted to create vacant spaces, it will cause the difference value histogram changing significantly and draw the attention of steganalyzers. So, this paper proposed a new idea for RDH based on the difference value and with statistical features maintained (SFM) with simple implementation and high scalability, we embed the secret messages by keeping the difference values that need to be modified in the original range, and the other difference values would not be shifted. In addition, we need the original difference values as the key to extract the secret messages. In order to expand the embedding capacity further, we designed two algorithms that embed message in two different difference values and four different difference values, and these two methods are named SFM_A and SFM_B respectively. SFM_B can support greater amount of embedded message than SFM_A, but brings greater changes to the original image, which could lead to the decline of PSNR and SSIM. The experimental results show that through our method, the histogram of difference values is well maintained, and the degree of distortion of the image is improved at the same time.

Now-a-days many unauthorized users try to get the protected information and therefore it is necessary to secure our data. Besides, there are also scenarios that data hiding needs to be done in the encrypted domain or combined with the encryption, especially in the age of big data and cloud computing. In previous method first encryption was done and then the room was vacated to hide the data. But there were some errors in data extraction and image recovery. In this paper a novel technique is proposed termed as reversible data hiding (RDH) by which we can hide our data. RDH is applied to encrypted images by which we can properly recover our data and the cover image. The hidden data can be in the form of text or image. In the proposed method we first vacate room to hide data and after encrypting image using certain encryption key the data hider reversibly hides the data whether text or image using data hiding key. In previous method only text was hidden. But the proposed paper is extended to hide the image also in the same reserved space which was used to hide text. The hidden image is also recovered without any errors. The results given in the paper proves that we can exactly recover the hidden data, hidden image and the cover image as and when needed by the receiver. The traditional RDH algorithm used is histogram shifting. By this method we can properly recover hidden data and cover image without any errors. Large amount of data can be hidden as compared to previous method.

A dual image based reversible data hiding (RDH) method using pixel value difference (PVD) and exploiting modification direction (EMD) are introduced in this paper. First enlarge the cover image size and then select a pair of pixel for data embedding. We embed 4 bits secret message through pixel value difference and 3 bits secret message using embedding function of exploiting modification direction within a pixel pair. We obtain two modified pairs of pixel which contain 7 bits secret message. We then distribute these pixel pairs among dual stego images depending on a shared secret key (K). The recipient successfully obtain secret message and retrieve original image using same shared secret key. We compared our proposed method with other existing techniques and find out averagely good results with respective to payload.

Reversible Data Hiding (RDH) should be secured as per requirements to protect content in open environments such as the cloud and internet. Integrity and undetectability of steganographic images are amongst the main concerns in any RDH scheme. As steganographic encryption using linear prediction over bit-planes is challenging, so the security and embedding capacity of the existing RDH techniques could not be adequate. Therefore, a new steganographic technique is proposed which provides better security, higher embedding capacity and visual quality to the RDH scheme. In this technique, the cover image is divided into n-bit planes (nBPs) and linear prediction is applied to it. Next, the histogram of the residual nBPs image is taken, and secret data bits are encrypted using the RC4 cryptographic algorithm. To embed the encrypted secret data bits, the histogram shifting process is applied. This is achieved by using peak and zero pairs of residual nBPs images. This scheme provides security to the cover image and hidden data. The proposed RDH scheme is capable of extracting the embedded secret data accurately and recovering the original cover or residual nBPs image.

Recently, Li et al. proposed a data hiding method based on pixel value ordering (PVO) and prediction error expansion (PEE). In their method, maximum and minimum values were predicted and the pixel values were modified to embed secret data. Thereafter, many scholars have proposed improvisations to the original PVO method. In this paper, a Reversible data hiding (RDH) method is proposed where the secret data is dispersed into two layers using different modes of operations. The second layer changes the dividing mode, and the first and the second layers do not take duplicate blocks. Under a fixed embedding capacity, threshold value and block size are controlled, complex blocks are filtered and the secret data is hidden in smooth blocks. This paper also compares the effectiveness of four well-known PVO series methods, the latest PVO methods, difference expansion (DE) method and reduced difference expansion (RDE) method. Experimental results show that the proposed method reduces distortion in the image, thereby enhancing the visual symmetry/quality compared to previous state-of-the-art methods and increasing its high application value.

Reversible data hiding (RDH) technique allows the original cover to be lossless restored after the secret message is extracted, and high dynamic range (HDR) images are becoming more and more popular. We found that the existing RDH schemes for HDR image will cause serious stream expansion, which means that the storage size of the cover HDR image will expand. Noticing that we proposed a fast coding method named reverse-Golomb code for message embedding in all-zero cover to reduce the number of the alteration of pixel’s status, and thus reduce the stream expansion of cover HDR images. The experimental results show the superiority of our method.

Optimization-based region-of-interest reconstruction for X-ray computed tomography based on total variation and data derivative

Acquired 3D point cloud data, whether from active sensors directly or from stereo-matching algorithms indirectly, typically contain non-negligible noise. To address the point cloud denoising problem, we propose a local graph-based algorithm. Specifically, given a $k$ -nearest-neighbor graph of the 3D points, we first approximate it with a bipartite graph (independent sets of red and blue nodes) using a KL divergence criterion. For each partite of nodes (say red), we first define surface normal of each red node using 3D coordinates of neighboring blue nodes, so that red node normals n can be written as a linear function of red node coordinates p. We then formulate a convex optimization problem, with a quadratic fidelity term $\Vert \mathbf{p}-\mathbf{q}\Vert_{2}^{2}$ given noisy observed red coordinates q and a graph total variation (GTV) regularization term for surface normals of neighboring red nodes. We minimize the resulting $l_{2}-l_{1}$-norm using alternating direction method of multipliers (ADMM) and proximal gradient descent. The two partites of nodes are alternately optimized until convergence. Experimental results show that compared to state-of-the-art schemes with similar complexity, our proposed algorithm achieves the best overall denoising performance objectively and subjectively.

We propose a deep learning solution to the inverse problem of localizing sources of network diffusion. Invoking graph signal processing (GSP) fundamentals, the problem boils down to blind estimation of a diffusion filter and its sparse input signal encoding the source locations. While the observations are bilinear functions of the unknowns, a mild requirement on invertibility of the graph filter enables a convex reformulation that we solve via the alternating-direction method of multipliers (ADMM). We unroll and truncate the novel ADMM iterations, to arrive at a parameterized neural network architecture for Source Localization on Graphs (SLoG-Net), that we train in an end-to-end fashion using labeled data. This way we leverage inductive biases of a GSP model-based solution in a data-driven trainable parametric architecture, which is interpretable, parameter efficient, and offers controllable complexity during inference. Experiments with simulated data corroborate that SLoG-Net exhibits performance in par with the iterative ADMM baseline, while attaining significant (post-training) speedups.

In this paper, a reversible data hiding algorithm for audio files based on code division multiplexing (CDM) is proposed. In the scheme, the orthogonal spreading sequences are employed to carry the embedding data, and the host audio file can also be recovered completely after the secret message has been extracted accurately. At the same time, according the orthogonal character of the spreading sequences, the secret message can be embedded into the audio files repeatedly, and most elements of the spreading sequences are mutually canceled in the process of multi-level data embedding, which maintains the audio file in good auditory even at high data embedding capacity. Moreover, only the receiver who holds the same spreading sequences, as the sender can restore the embedded data, which enhance the security of the proposed scheme. As shown in the experimental results, the CDM based reversible data hiding (RDH) algorithm for audio files can achieve higher data embedding capacity at the same audio distortion compared with those state-of-the-art audio RDH algorithms.

Reversible data hiding (RDH) is crucial in modern data security, ensuring confidentiality and tamper-proofness in various industries like copyright protection, medical imaging, and digital forensics. As technology advances, RDH techniques become essential, but the trade-off between embedding capacity and visual quality must be heeded. In this paper, the relative correlation between the pixel's local complexity and its directional prediction error is employed to enhance an efficient RDH without using a location map. An embedding process based on multiple cumulative peak region localization (MCPRL) is proposed to hide information in the 3D-directional prediction error histogram with a lower local complexity value and avoid the underflow/overflow problems. The carrier image is divided into three color channels, and then each channel is split into two non-overlapping sets: blank and shadow. Two half-directional prediction errors (the blank set and the shadow set) are constructed to generate a full-directional prediction error for each color channel belonging to the host image. The local complexity value and directional prediction error are critical metrics in the proposed embedding process to improve security and robustness. By utilizing these metrics to construct a 3D stego-Blank Set, the 3D stego-shadow Set will be subsequently constructed using the 3D blank set. The proposed technique outperforms other state-of-the-art techniques in terms of embedding capacity, image quality, and robustness against attacks without an extra location map. The experimental results illustrate the effectiveness of the proposed method for various 3D RDH techniques.