【论文阅读】图像信息隐藏文章汇总（含代码）-Toy模板网

这篇具有很好参考价值的文章主要介绍了【论文阅读】图像信息隐藏文章汇总（含代码）。希望对大家有所帮助。如果存在错误或未考虑完全的地方，请大家不吝赐教，您也可以点击"举报违法"按钮提交疑问。

<Large-capacity Image Steganography Based on Invertible Neural Networks>CVPR2021;可逆网络ISN，大容量的实现是靠RGB通道的累加；无公开代码
<Multitask Identity-Aware Image Steganography via Minimax Optimization>IEEE Transactions on Image Processing2021;提出直接识别防止接收端泄密、其中恢复分支可选；主要涉及身份信息;有公开代码GitHub - jiabaocui/MIAIS
<HiNet: Deep Image Hiding by Invertible Network>ICCV2021;可逆网络HiNet；与ISN同期首次在图像隐藏运用可逆网络，未实现大容量以及串联分阶段隐藏，强调可逆网络的性能优势;有公开代码GitHub - TomTomTommi/HiNet: Official PyTorch implementation of "HiNet: Deep Image Hiding by Invertible Network" (ICCV 2021)
<Breaking Robust Data Hiding in Online Social Networks>IEEE Signal Processing Letters 2022;隐写分析；OSN背景（FUDAN）；不需要原始载体图像，直接输入隐写图像，输出质量略有提升的载体图像；三个模块分别：去除隐藏的秘密数据、增强图像质量和提高整体性能;无公开代码
<Image Generation Network for Covert Transmission in Online Social Network>ACMMM2022;图像隐写，OSN背景（FUDAN）；无覆盖式（没载体图），生成模块对抗模块提取模块噪声模块，输入秘密和目标表情，生成含有秘密的表情包人脸图;无工开代码
<Robust Invertible Image Steganography>CVPR2022;可逆网络强调鲁棒；鲁棒靠载体增强模块实现（消除收到的载密图像噪声以及JPEG压缩失真的影响）；无公开代码
<Image Disentanglement Autoencoder for Steganography Without Embedding>CVPR2022;无嵌入生成隐写解纠缠自动编码器，直接生成stego，不需要载体图像，分成结构和纹理两种表示，并用结构纹理的多种组合形成多种隐写图像，再提取结构信息恢复秘密;有公开代码GitHub - Lemok00/IDEAS: Official pytorch implementation of paper "Image Disentanglement Autoencoder for Steganography without Embedding" (CVPR2022).
<Fixed Neural Network Steganography: Train The Images, Not The Network>ICLR2022;FNNS 是在steganoGAN上的改进；主要强调降低解码错误率：在3bpp的情况下实现精确的0%错误率;有代码GitHub - varshakishore/FNNS
<Generative Steganography Network>ACMMM2022;无嵌入生成隐写：根据输入对（潜在向量和噪声/秘密信息）不同，可选择生成cover还是stego;无公开代码
<StegGAN: hiding image within image using conditional generative adversarial networks>Multimedia Tools and Applications2022;有公开代码GitHub - brijeshiitg/StegGAN
<Large-capacity and Flexible Video Steganography via Invertible Neural Network>CVPR2023;视频隐写；可以在一个载体视频中隐藏最多7个秘密视频；提出一种秘钥可控方案；多视频隐藏可变数量方案;有代码Large-capacity and Flexible Video Steganography via Invertible Neural Network | Papers With Code
<DeepMIH: Deep Invertible Network for Multiple Image Hiding>TPAMI2023;可逆网络hinet升级版，强调串联多图像隐藏，有imp模块引导第二次嵌入；有感知损失；做了频域子带分离试验，结论是高频子带适合隐藏信息;有代码，和hinet一个作者主页
<SteganoGAN: High Capacity Image Steganography with GANs>2019;Jupiter；解码器编码器判别器;有代码SteganoGAN: High Capacity Image Steganography with GANs | Papers With Code
<HiDDeN: Hiding Data With Deep Networks>2018ECCV;代码：GitHub - ando-khachatryan/HiDDeN: Pytorch implementation of paper "HiDDeN: Hiding Data With Deep Networks" by Jiren Zhu, Russell Kaplan, Justin Johnson, and Li Fei-Fei
<RoSteALS: Robust Steganography using Autoencoder Latent Space>2023CVPR;针对现有方法因为保证图像质量和抗扰动鲁棒性导致的训练复杂的问题，提出了一种冻结预训练编码器的训练简单的轻量级网络。GitHub - TuBui/RoSteALS: RoSteALS: Robust Steganography using Autoencoder Latent Space文章来源地址https://www.toymoban.com/news/detail-845810.html
<Towards Robust Data Hiding Against (JPEG) Compression: A Pseudo-Differentiable Deep Learning Approach>2020;针对(jpeg)压缩的稳健数据隐藏:一种伪可微深度学习方法mikolez/Robust_JPEG · GitHub
<Robust Image Steganography: Hiding Messages in Frequency Coefficients>2023AAAI;基于INN的鲁棒隐写方法；尤其是针对JPEG压缩具有良好鲁棒性；秘密信息形式为二进制；无公开代码
<MBRS：Enhancing Robustness of DNN-based Watermarking by Mini-Batch of Real and Simulated JPEG>MM '21;对于不同的小批量，随机选择真实JPEG、模拟JPEG和无噪声层中的一个作为噪声层；鲁棒方法；有代码jzyustc/MBRS: This is the source code of paper MBRS : Enhancing Robustness of DNN-based Watermarking by Mini-Batch of Real and Simulated JPEG Compression, which is received by ACM MM' 21. (github.com)