High-Capacity Convolutional Video Steganography with Temporal Residual Modeling

Abstract

Steganography represents the art of unobtrusively concealing a secrete message within some cover data. The key scope of this work is about visual steganography techniques that hide a full-sized color image / video within another. A majority of existing works are devoted to the image case, where both secret and cover data are images. We empirically validate that image steganography model does not naturally extend to the video case (i.e., hiding a video into another video), mainly because it completely ignores the temporal redundancy within consecutive video frames. Our work proposes a novel solution to the problem of video steganography. The technical contributions are two-fold: first, the residual between two consecutive frames tends to zero at most pixels. Hiding such highly-sparse data is significantly easier than hiding the original frames. Motivated by this fact, we propose to explicitly consider inter-frame residuals rather than blindly applying image steganography model on every video frame. Specifically, our model contains two branches, one of which is specially designed for hiding inter-frame difference into a cover video frame and the other instead hides the original secret frame. A simple thresholding method determines which branch a secret video frame shall choose. When revealing the concealed secret video, two decoders are devised, revealing difference or frame respectively. Second, we develop the model based on deep convolutional neural networks, which is the first of its kind in the literature of video steganography. In experiments, comprehensive evaluations are conducted to compare our model with both classic least significant bit (LSB) method and pure image steganography models. All results strongly suggest that the proposed model enjoys advantages over previous methods. We also carefully investigate key factors in the success of our deep video steganography model. The first two authors contribute equally. Yadong Mu is the corresponding author of this work. cover secret Alice container Eve Bob decoded secret contain secret message? Figure 1: The full scheme of steganography. See main text for more explanation.

Payload location for JPEG image steganography based on co-frequency sub-image filtering

“…with audio, video, and other media types as covers. 32 It may be a potential idea to search similar cover media 33 and apply the data learning 34 to recognize the stego positions.…”

A Comprehensive Tutorial and Survey of Applications of Deep Learning for Cyber Security

“…With the aim to increase the payload capacity without changing the appearance of cover image in a larger term, CNN based method is employed [107]. CNN based image steganography is mapped into video steganography by [108]. DNN based secret information removal is studied by [109].…”

Advanced Intelligent Data Hiding Using Video Stego and Convolutional Neural Networks

“…The first advantage, is that the convent would is a good idea with regards to patterns of original images and are going to be capable of making decisions on what areas are redundant. As a result, more pixels will be hidden there (8). Through saving space on redundant areas, the hidden information amount will be increased since the weights and architecture could be randomized.…”

“…Several of these researches utilized CNN to choose what is the best LSBs to substitution an image with the binary to embed a text message. Others, (15)(16)(17), have utilized CNN to determine which is the desired bits to extract from the cover images. In this work, the neural network is used to determines where it can place the secret data and how exactly to encode it effectively; the hidden message can be disappeared through the entire bits in the image.…”