Content-Adaptive Steganography by Minimizing Statistical Detectability

By Vahid Sedighi, Rémi Cogranne, Jessica J. Fridrich. IEEE Transactions on Information Forensics and Security, 2016

526 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/32035db46a6ee6e533b210c139b6f10f919f3fdc

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

1. Lu Yuan
2. Dongdong Chen
3. Bin Li
4. Nenghai Yu
5. Jiwu Huang
6. Jing Liao
7. Gang Hua
8. Mo Chen

Citations in the corpus

• Automatic Steganographic Distortion Learning Using a Generative Adversarial Network (2017, methodology): “= (ρi,j )H×W is heuristically designed in a content-adaptive manner in all of the state-of-the-art spatial steganographic schemes, including HUGO [20], S-UNIWARD [21], MiPOD [22], and HILL [23].”
• Deep Residual Network for Steganalysis of Digital Images (2019, methodology): “Embedding modifications introduced by spatial-domain embedding methods that minimize an additive distortion, such as WOW [28], HILL [41], S-UNIWARD [31], and MiPOD [47] are largely uncorrelated, while changes to quantized DCT coefficients in JPEG image steganography lead to a stego signal with signi”
• A New Payload Partition Strategy in Color Image Steganography (2020, methodology): “Later Sedighi et al. [15] derived a closed form expression for the detector of content-adaptive LSB matching within the selected model, and then designed steganography based on minimizing the power of the optimal detector (MiPOD).”
• Depth-Wise Separable Convolutions and Multi-Level Pooling for an Efficient Spatial CNN-Based Steganalysis (2020, unknown): “istics [1]–[3], [50], [51], [57].”
• Deep residual learning for image steganalysis (2018, methodology): “4 bpp Steganography SRM maxSRMd2 DRN WOW 20.1 % 15.2 % 4.3 % S-UNIWARD 20.3 % 18.8 % 6.3 % HILL 24.2 % 21.6 % 10.4 % MiPOD 22.1 % 20.4 % 4.9 % high embedding probability.”
• E2Style: Improve the Efficiency and Effectiveness of StyleGAN Inversion (2021, background): “In contrast to traditional steganography [67], [68], [69], which embeds secret messages with small capacity, deep hiding aims at large capacity hidden messages, such as hiding a secret image with 1024 × 1024 resolution into a carrier image.”