Steganalysis using learned denoising kernels
- PDF / 1,313,821 Bytes
- 15 Pages / 439.642 x 666.49 pts Page_size
- 102 Downloads / 210 Views
Steganalysis using learned denoising kernels Brijesh Singh1
· Mohit Chhajed1 · Arijit Sur1 · Pinaki Mitra1
Received: 15 August 2019 / Revised: 11 September 2020 / Accepted: 22 September 2020 / © Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract Steganalysis is the science for detecting steganographic traces in innocent-looking digital media like images, videos, etc. In recent literature, it has been observed that state-ofthe-art image steganographic techniques such as S-UNIWARD, HUGO, WOW, etc. still remain undetected even with considerable embedding payload. Recently, the deep learning framework has been hugely successful in different computer vision applications like object detection, image classification, event detection, etc. Some recent deep learning-based works also show promising results for image steganalysis and have opened a new avenue for research. The current literature reveals that the steganalytic detector becomes more precise if trained on the residual error (embedding noise) domain. To get an accurate noise residual, it is required to predict the cover image precisely from the corresponding stego image. In this work, a denoising kernel has been learned to obtain a more precise noise residual. After that, a CNN based steganalytic detector is devised, which is trained using the noise residual to get a more precise detection. Experimental results show that the proposed scheme outperforms the state-of-the-art steganalysis schemes against the state-of-the-art steganographic approaches. Keywords Steganalysis · Spatial domain steganalysis · Denoising kernels
Brijesh Singh
[email protected] Mohit Chhajed [email protected] Arijit Sur [email protected] Pinaki Mitra [email protected] 1
Department of CSE, Indian Institute of Technology, Guwahati, India
Multimedia Tools and Applications
1 Introduction Steganography is an ancient art of covert writing. With the growing popularity of digital communication, digital steganography has become an emerging research topic. Digital steganography includes– steganography in images [29], text [32], videos [33], etc. One of the major challenges of digital image steganography is maintaining its statistical undetectability, which means, the embedding noise should not be detected by any statistical detectors more than random guessing. Some of the state-of-the-art steganographic schemes are Spatial Universal Wavelet Relative Distortion (S-UNIWARD) [14], Highly Undetectable stego (HUGO) [28], Wavelet Obtained Weights (WOW) [12], MiPOD [22, 23, 34], etc. On the other side, steganalysis is the science of investigating the traces of hidden data within an innocent-looking cover media. In the literature, image steganalysis has two major modules. The first module is feature engineering, where an image is approximated as a statistical model to find the distinguishing features due to steganographic embedding, and the next is building a steganalytic classifier by training machine learning algorithms like SVM, LDA, using these features. These f
Data Loading...
