Open Access

ARTICLE

A Deep Learning Driven Feature Based Steganalysis Approach

Yuchen Li¹, Baohong Ling^1,2,*, Donghui Hu¹, Shuli Zheng¹, Guoan Zhang³

1 College of Computer Science and Information Engineering, Hefei University of Technology, Hefei, 230009, China
2 College of Information Engineering, Anhui Broadcasting Movie and Television College, Hefei, 230011, China
3 Department of Informatics, Faculty of Natural & Mathematical Sciences, King’s College London, London, WC2R2LS, UK

* Corresponding Author: Baohong Ling. Email:

Intelligent Automation & Soft Computing 2023, 37(2), 2213-2225. https://doi.org/10.32604/iasc.2023.029983

Received 16 March 2022; Accepted 25 May 2022; Issue published 21 June 2023

Abstract

The goal of steganalysis is to detect whether the cover carries the secret information which is embedded by steganographic algorithms. The traditional steganalysis detector is trained on the stego images created by a certain type of steganographic algorithm, whose detection performance drops rapidly when it is applied to detect another type of steganographic algorithm. This phenomenon is called as steganographic algorithm mismatch in steganalysis. To resolve this problem, we propose a deep learning driven feature-based approach. An advanced steganalysis neural network is used to extract steganographic features, different pairs of training images embedded with steganographic algorithms can obtain diverse features of each algorithm. Then a multi-classifier implemented as lightgbm is used to predict the matching algorithm. Experimental results on four types of JPEG steganographic algorithms prove that the proposed method can improve the detection accuracy in the scenario of steganographic algorithm mismatch.

---

Keywords

---