Recent work has shown that deep neural networks are highly sensitive to tiny perturbations of input images, giving rise to adversarial examples Though this property is usually considered a weakness of learned models, we explore whether it can be beneficial We find that neural networks can learn to use invisible perturbations to encode a rich amount of useful information In fact, one can exploit this capability for the task of data hiding We jointly train encoder and decoder networks, where given an input message and cover image, the encoder produces a visually indistinguishable encoded image, from which the decoder can recover the original message We show that these encodings are competitive with existing data hiding algorithms, and further that they can be made robust to noise: our models learn to reconstruct hidden information in an encoded image despite the presence of Gaussian blurring, pixel-wise dropout, cropping, and JPEG compression Even though JPEG is non-differentiable, we show that a robust model can be trained using differentiable approximations Finally, we demonstrate that adversarial training improves the visual quality of encoded images

/pdf/hidden-hiding-data-with-deep-networks-1kw8ewaemh.pdf

HiDDeN: Hiding Data With Deep Networks

Reversible data hiding in encrypted images has attracted considerable attention from the communities of privacy security and protection. The success of the previous methods in this area has shown that a superior performance can be achieved by exploiting the redundancy within the image. Specifically, because the pixels in the local structures (like patches or regions) have a strong similarity, they can be heavily compressed, thus resulting in a large hiding room. In this paper, to better explore the correlation between neighbor pixels, we propose to consider the patch-level sparse representation when hiding the secret data. The widely used sparse coding technique has demonstrated that a patch can be linearly represented by some atoms in an over-complete dictionary. As the sparse coding is an approximation solution, the leading residual errors are encoded and self-embedded within the cover image. Furthermore, the learned dictionary is also embedded into the encrypted image. Thanks to the powerful representation of sparse coding, a large vacated room can be achieved, and thus the data hider can embed more secret messages in the encrypted image. Extensive experiments demonstrate that the proposed method significantly outperforms the state-of-the-art methods in terms of the embedding rate and the image quality.

High Capacity Reversible Data Hiding in Encrypted Images by Patch-Level Sparse Representation

The proliferation of digitized media due to the rapid growth of networked multimedia systems, has created an urgent need for copyright enforcement technologies that can protect copyright ownership of multimedia objects. Digital image watermarking is one such technology that has been developed to protect digital images from illegal manipulations. In particular, digital image watermarking algorithms which are based on the discrete wavelet transform have been widely recognized to be more prevalent than others. This is due to the wavelets' excellent spatial localization, frequency spread, and multi-resolution characteristics, which are similar to the theoretical models of the human visual system. In this paper, we describe an imperceptible and a robust combined DWT-DCT digital image watermarking algorithm. The algorithm watermarks a given digital image using a combination of the Discrete Wavelet Transform (DWT) and the Discrete Cosine Transform (DCT). Performance evaluation results show that combining the two transforms improved the performance of the watermarking algorithms that are based solely on the DWT transform.

Combined DWT-DCT Digital Image Watermarking

This paper proposes a blind watermarking algorithm based on the significant difference of wavelet coefficient quantization for copyright protection. Every seven nonoverlap wavelet coefficients of the host image are grouped into a block. The largest two coefficients in a block are called significant coefficients in this paper and their difference is called significant difference. We quantized the local maximum wavelet coefficient in a block by comparing the significant difference value in a block with the average significant difference value in all blocks. The maximum wavelet coefficients are so quantized that their significant difference between watermark bit 0 and watermark bit 1 exhibits a large energy difference which can be used for watermark extraction. During the extraction, an adaptive threshold value is designed to extract the watermark from the watermarked image under different attacks. We compare the adaptive threshold value to the significant difference which was quantized in a block to determine the watermark bit. The experimental results show that the proposed method is quite effective against JPEG compression, low-pass filtering, and Gaussian noise; the PSNR value of a watermarked image is greater than 40 dB.

An Efficient Watermarking Method Based on Significant Difference of Wavelet Coefficient Quantization

https://ieeexplore.ieee.org/iel5/6046/4563439/04536064.pdf

An Efficient Watermarking Method Based on Significant Difference of

In this paper, an image accreditation technique by embedding digital watermarks in images is proposed. The proposed method for the digital watermarking is based on the wavelet transform. This is unlike most previous work, which used a random number of a sequence of bits as a watermark and where the watermark can only be detected by comparing an experimental threshold value to determine whether a sequence of random signals is the watermark. The proposed approach embeds a watermark with visual recognizable patterns, such as binary, gray, or color image in images by modifying the frequency part of the images. In the proposed approach, an original image is decomposed into wavelet coefficients. Then, multi-energy watermarking scheme based on the qualified significant wavelet tree (QSWT) is used to achieve the robustness of the watermarking. Unlike other watermarking techniques that use a single casting energy, QSWT adopts adaptive casting energy in different resolutions. The performance of the proposed watermarking is robust to a variety of signal distortions, such as JPEG, image cropping, sharpening, median filtering, and incorporating attacks.

Hiding digital watermarks using multiresolution wavelet transform

Digital watermarking has been presented to protect the copyright of electronic multimedia data. The previous proposed watermarking algorithms mostly embed sequences of random numbers or binary images as the watermarks. In this paper, an image accreditation technique by embedding digital gray-level image watermarks in images is proposed. In the proposed method, the host image and the watermark are decomposed into wavelet coefficients. The contextual energies of the host-image wavelet coefficients are computed to select embedding coefficients. The watermark wavelet coefficients are sorted and then embedded in the selected host coefficients. The experiments show that the proposed method provides extra robustness against JPEG-compression, image-processing, and even composite attacks compared to the traditional embedding methods. Moreover, the proposed method has no need of the original image to extract the embedded watermarks. The proposed method facilitates electronic commerce application developers to select adequate digital watermarking techniques for their development systems.

Perceptual Digital Watermarking for Image Authentication in Electronic Commerce

Wavelet image decomposition generates a hierarchical data structure to represent an image. Recently, a new class of image compression algorithms has been developed for exploiting dependencies between the hierarchical wavelet coefficients using zerotrees. This paper deals with a fuzzy inference filter for image entropy coding by choosing significant coefficients and zerotree roots in the higher frequency wavelet subbands. Moreover, an adaptive quantization is proposed to improve the coding performance. Evaluating with the standard images, the proposed approaches are comparable or superior to most state-of-the-art coders. Based on the fuzzy energy judgment, the proposed approaches can achieve an excellent performance on the combination applications of image compression and watermarking.

Image subband coding using fuzzy inference and adaptive quantization

In this paper, an image accreditation technique by embedding digital watermarks in images is proposed. The proposed method for the digital watermarking is based on the wavelet transform. This is unlike most previous work, which used a random number of a sequence of bits as a watermark and where the watermark can only be detected by comparing an experimental threshold value to determine whether a sequence of random signals is the watermark. The proposed approach embeds a watermark with visual recognizable patterns, such as binary, gray, or color image in images by modifying the frequency part of the images. In the proposed approach, an original image is decom- posed into wavelet coefficients. Then, multi-energy watermarking scheme based on the qualified significant wavelet tree (QSWT) is used to achieve the robustness of the watermarking. Unlike other watermarking techniques that use a single casting energy, QSWT adopts adaptive casting energy in different resolutions. The performance of the proposed watermarking is robust to a variety of signal distortions, such as JPEG, image cropping, sharpening, median filtering, and incorporating attacks.

Hiding Digital Watermarks Using Multiresolution

A class of image compression algorithms has been developed for exploiting dependencies between the hierarchical wavelet coefficients using zerotrees The paper deals with a fuzzy inference filter for image entropy coding by choosing significant coefficients and zerotree roots in the higher frequency wavelet subbands Moreover, an adaptive quantization is proposed to improve the coding performance Evaluating with the popular images, the proposed approach is comparable or superior to most state-of-the-art coders

Ming-Shing Hsieh

Papers

Hiding digital watermarks using multiresolution wavelet transform

Perceptual Digital Watermarking for Image Authentication in Electronic Commerce

Image subband coding using fuzzy inference and adaptive quantization

Hiding Digital Watermarks Using Multiresolution

Wavelet-based image coding using fuzzy inference and adaptive quantization