Ensemble Deep Learning Features for Real-World Image Steganalysis = Ensemble Deep Learning Features for Real-World Image Steganalysis

다국어 초록 (Multilingual Abstract)

The Alaska competition provides an opportunity to study the practical problems of real-world steganalysis. Participants are required to solve steganalysis involving various embedding schemes, inconsistency JPEG Quality Factor and various processing pipelines. In this paper, we propose a method to ensemble multiple deep learning steganalyzers. We select SRNet and RESDET as our base models. Then we design a three-layers model ensemble network to fuse these base models and output the final prediction. By separating the three colors channels for base model training and feature replacement strategy instead of simply merging features, the performance of the model ensemble is greatly improved. The proposed method won second place in the Alaska 1 competition in the end.

참고문헌 (Reference)

1 L. Guo, "Using Statistical Image Model for JPEG Steganography : Uniform Embedding Revisited" 10 (10): 2669-2680, 2015

2 V. Holub, "Universal distortion function for steganography in an arbitrary domain" 2014 (2014): 1-13, 2014

3 L. Guo, "Uniform Embedding for Efficient JPEG Steganography" 9 (9): 814-825, 2014

4 C. F. Tsang, "Steganalyzing images of arbitrary size with CNNs" (7) : 121-1-121-8, 2018

5 X. Song, "Steganalysis of adaptive JPEG steganography using 2D Gabor filters" 15-23, 2015

6 X. S. Huang, "Steganalysis of Adaptive JPEG Steganography Based on ResDet" 2018

7 Q. Gibouloto, "Steganalysis into the wild : How to define a source?" (7) : 318-1-318-112, 2018

8 F. Jessica, "Statistically undetectable jpeg steganography : dead ends challenges, and opportunities" 3-14, 2007

9 I. Loshchilov, "SGDR: Stochastic gradient descent with warm restarts"

10 V. Holub, "Phase-aware projection model for steganalysis of JPEG images" 2015

11 V. Holub, "Low-complexity features for JPEG steganalysis using undecimated DCT" 10 (10): 219-228, 2015

12 J. S. Zeng, "Largescale JPEG image steganalysis using hybrid deep-learning framework" 13 (13): 1200-1214, 2018

13 M. Chen, "JPEG-phase-aware convolutional neural network for steganalysis of JPEG images" 75-84, 2017

14 Beijing Chen, "High-Capacity Robust Image Steganography via Adversarial Network" 한국인터넷정보학회 14 (14): 366-381, 2020

15 J. Yang, "Feature fusion : parallel strategy vs. serial strategy" 36 (36): 1369-1381, 2003

16 G. Huang, "Densely Connected Convolutional Networks" 2261-2269, 2017

17 M. Boroumand, "Deep residual network for steganalysis of digital images" 14 (14): 1181-1193, 2018

18 G. S. Xu, "Deep convolutional neural network to detect J-UNIWARD" 67-73, 2017

19 S. Ioffe, "Batch normalization : Accelerating deep network training by reducing internal covariate shift"

20 W. K. You, "A Siamese CNN for Image Steganalysis" 16 : 291-306, 2020