Deep learning for steganalysis via convolutional neural networks

By Yinlong Qian, Jing Dong, Wei Wang, Tien-Ping Tan. Electronic imaging, 2015

544 citations recorded by the source corpus.

Current work on steganalysis for digital images is focused on the construction of complex handcrafted features. This paper proposes a new paradigm for steganalysis to learn features automatically via deep learning models. We novelly propose a customized Convolutional Neural Network for steganalysis. The proposed model can capture the complex dependencies that are useful for steganalysis. Compared with existing schemes, this model can automatically learn feature representations with several convolutional layers. The feature extraction and classification steps are unified under a single architecture, which means the guidance of classification can be used during the feature extraction step. We demonstrate the effectiveness of the proposed model on three state-of-theart spatial domain steganographic algorithms - HUGO, WOW, and S-UNIWARD. Compared to the Spatial Rich Model (SRM), our model achieves comparable performance on BOSSbase and the realistic and large ImageNet database.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/f7100e9fd9b87c7d8cff9fea986254afadb392ea

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

Citations in the corpus

Depth-Wise Separable Convolutions and Multi-Level Pooling for an Efficient Spatial CNN-Based Steganalysis (2020, methodology): “[25] proposed a steganalysis model using standard CNN architecture”
Deep residual learning for image steganalysis (2018, methodology): “in [26] proposed a CNN model by using the Gaussian activation function and average pooling.”
Recasting Residual-based Local Descriptors as Convolutional Neural Networks: an Application to Image Forgery Detection (2017, methodology): “Taking advantage of the lesson learnt from SPAM/SRM features, constrained CNN architectures have been proposed both for steganalysis [23] and manipulation detection [2], where the rst convolutional layer is forced to perform a high-pass ltering.”
Deep Learning for Deepfakes Creation and Detection (2019, background): “This increases the pixel level statistical similarity between real images and fake images and requires the forensic classiﬁer to learn more intrinsic and meaningful features, which has better generalization capability than previous image forensics methods [51,52] or image steganalysis networks [53].”
Yedroudj-Net: An Efficient CNN for Spatial Steganalysis (2018, unknown): “Inspired by the benefit of diversity [12], and similarly to [17], we use the 30-basic high-pass filters from SRM [12], instead of using only one filter such as [2, 3, 5], in order to pre-process the input image.”
Deep Learning Hierarchical Representations for Image Steganalysis (2017, methodology): “’s work [11], only the “square 5 × 5” filter in SRM was used to initialize their first layer in the GNCNN model.”
Noiseprint: A CNN-Based Camera Model Fingerprint (2018, background): “tures, inspired again by work in steganalysis [45], all focus on suppressing the scene content, forcing the network to”
Hybrid LSTM and Encoder–Decoder Architecture for Detection of Image Forgeries (2019, background): “In [75], the authors propose Gaussian-Neuron CNN (GNCNN) for steganal-ysis.”
Deep Residual Network for Steganalysis of Digital Images (2019, background): “[45] the authors described a neural network steganalyzer with a Gaussian activation function equipped with a fixed preprocessing high-pass KV filter [39, eq.”
Media Forensics and DeepFakes: An Overview (2020, methodology): “by a similar solution used in [137] for steganalysis, a CNN is proposed with a constrained first layer that performs high-pass filtering of the image, that suppress the scene content and allow to work on residuals.”
Deep learning for deepfakes creation and detection: A survey (2019, background): “similarity between real images and fake images and requires the forensic classifier to learn more intrinsic and meaningful features, which has better generalization capability than previous image forensics methods [61], [62] or image steganalysis networks [63].”