Latest Trends in Deep Learning Techniques for Image Steganography

Kumar, Vijay; Sharma, Sahil; Kumar, C. Santhosh; Sahu, Aditya Kumar

doi:10.4018/ijdcf.318666

Cited by 13 publications

(6 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One of the results of the proposed algorithm for the selection of the stego area, based on the homogenization of the parts of the carrier, can be the detection of typical shapes, objects and contours. As algorithms based on deep learning techniques presented in [26] have the same goal, one of the directions of further research could be to combine these techniques.…”

Section: Discussionmentioning

confidence: 99%

Steganographic Method in Selected Areas of the Stego-Carrier in the Spatial Domain

Milosav¹,

Milosavljević

Banjac³

2023

Symmetry

View full text Add to dashboard Cite

The main goal of this paper is the proposal of a key-based steganographic system in which the ratio of capacity and image quality metrics that represents the stego object while reducing the detectability of hidden content was improved. The main contribution of the proposed steganographic system is a new algorithm for selecting stego areas. The area selection algorithm is based on clustering the pixels of the cover object into a predetermined number of clusters. The goal of this selection of areas (clusters) is to group as many homogeneous parts of the image as possible in order to cover these areas with as few rectangular shapes as possible. Since the data on the defined rectangles represent the key of the system, the capacity of the additional secret channel is minimized in this way. On the obtained stego-carriers, an embedding of test random content is performed in order to estimate its detectability. By combining the proposed area selection method with the Minimal Decimal Difference steganographic method, a system was created with an optimal trade-off between detectability of secret content, quality and capacity of the carrier, and the length of the stego-key. Finally, a comparison of the obtained results with relevant adaptive steganographic methods is presented. The proposed concept obtains its verification in one practical system for secure file transfer of controlled cryptographic strength.

show abstract

Section: Discussionmentioning

confidence: 99%

Steganographic Method in Selected Areas of the Stego-Carrier in the Spatial Domain

Milosav¹,

Milosavljević

Banjac³

2023

Symmetry

View full text Add to dashboard Cite

show abstract

“…In addition, this paper investigates whether these multiple-image steganographic schemes can be detected by modern single-image steganographic analysis tools. An overview of current developments in deep learningbased image steganography is presented in this paper [27]. Using deep learning techniques to steganographer images, the authors present various works and provide some notes on how the different methods work.…”

Section: Related Workmentioning

confidence: 99%

A Fast and Low Distortion Image Steganography Framework Based on Nature-Inspired Optimizers

Fadel,

Said,

Hagras

et al. 2023

IEEE Access

View full text Add to dashboard Cite

Nowadays, secret information is susceptible to numerous hacks while being shared through traditional correspondence channels. Unlike other methods of securing data, steganography offers the most comprehensive security. Several algorithms have been developed to determine the optimal embedding capacity of every pixel to enhance steganography techniques. Concealing secret bits is not the only challenging aspect of this operation; improving the embedding capacity of secret data is also important, as is maintaining an acceptable visual quality after concealment. A new framework is proposed for concealing a secret text in the spatial domain by partitioning the host image into some non-overlapping blocks and classifying each block as edged or smooth using the Whale Optimization Algorithm (WOA). Different WOA objective functions can be utilized to determine each pixel's embedding capacity based on its intensity. For smooth blocks, the fitness function (with low objective values) is used, and for edge blocks, the cost function (with high objective values) is used instead. Additionally, several combinations of scanning order and starting point for each block in the host image are determined to reduce embedding distortion. WOA can be used to determine the best combination of them. Minimum replacement error (MRE) and compensated grayscale changes (CGSC) modules further reduce embedded distortion and artifacts. Accordingly, the recommended technique produces a peak signal-to-noise ratio (PSNR) of 62.44 dB, compared to 43.15 dB using the benchmark algorithm Multiple Pixel-value Adjustment (MPA) with an encoding function algorithm. Due to its superior performance, the recommended framework outperforms existing frameworks.

show abstract

“…However, this requires extra gradients from the face restoration model, so it is not easy to extend across models. Referring to the training procedure of image steganography models [43]- [46], we introduce a residual decoder D η (•) to recover the trigger image from the residual between the poisoned and benign image. The SF-I-Net ensures the poisoned image is indistinguishable from the benign image, while the decoder ensures that the poisoned image contains trigger information.…”

Section: ) Training Strategymentioning

confidence: 99%

FIBA: Frequency-Injection based Backdoor Attack in Medical Image Analysis

Feng

Zhang

et al. 2022

2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

View full text Add to dashboard Cite

Deep learning-based face restoration models, increasingly prevalent in smart devices, have become targets for sophisticated backdoor attacks. These attacks, through subtle trigger injection into input face images, can lead to unexpected restoration outcomes. Unlike conventional methods focused on classification tasks, our approach introduces a unique degradation objective tailored for attacking restoration models. Moreover, we propose the Adaptive Selective Frequency Injection Backdoor Attack (AS-FIBA) framework, employing a neural network for input-specific trigger generation in the frequency domain, seamlessly blending triggers with benign images. This results in imperceptible yet effective attacks, guiding restoration predictions towards subtly degraded outputs rather than conspicuous targets. Extensive experiments demonstrate the efficacy of the degradation objective on state-of-the-art face restoration models. Additionally, it is notable that AS-FIBA can insert effective backdoors that are more imperceptible than existing backdoor attack methods, including WaNet, ISSBA, and FIBA.

show abstract

Latest Trends in Deep Learning Techniques for Image Steganography

Cited by 13 publications

References 50 publications

Steganographic Method in Selected Areas of the Stego-Carrier in the Spatial Domain

Steganographic Method in Selected Areas of the Stego-Carrier in the Spatial Domain

A Fast and Low Distortion Image Steganography Framework Based on Nature-Inspired Optimizers

FIBA: Frequency-Injection based Backdoor Attack in Medical Image Analysis

Contact Info

Product

Resources

About