091635Aa/art-asset-protection-system

# 美术资产保护系统 (Art Asset Protection System) [中文](#中文说明) | [English](#english-description) ## 中文说明 一套完整的美术资产 AI 防护与版权保护解决方案，提供多层防护算法、效果验证、盲水印、图像指纹比对等功能。 ### 功能特性 #### 五层防护算法 - **Fawkes 特征防护** - 干扰 AI 特征提取，保护图像特征 - **Glaze 画风防护** - 保护艺术风格不被 AI 模仿 - **Foolbox 对抗噪点** - 添加对抗噪点干扰 AI 识别 - **OpenStego 隐写误导** - LSB 隐写注入误导数据 - **Blind Watermark 盲水印** - DWT-DCT-SVD 不可见数字水印 #### 效果验证系统 - 特征向量欧氏距离验证 - 画风相似度验证 - 结构识别准确率验证 - 隐写数据错误率验证 - 可视化验证报告生成 #### 盲水印功能 (V2/V3) - 文本水印嵌入与提取 - 图像水印嵌入与提取 - 密码加密保护 #### ChainGuard 链智护 (V3) - 单图 2048 位特征指纹提取 - 双图指纹相似度比对 - 版权盗版快速识别 #### MaskSource 模型溯源 (V3) - AI 模型水印保护介绍 - 双盾双溯源技术体系 ### 版本说明 | 版本 | 防护工具 | 效果验证 | 盲水印 | ChainGuard | MaskSource | |------|---------|---------|--------|------------|------------| | V1 | [x] | [x] | [ ] | [ ] | [ ] | | V2 | [x] | [x] | [x] | [ ] | [ ] | | V3 | [x] | [x] | [x] | [x] | [x] | ### 快速开始 #### 环境要求 - Python 3.8+ - 内存: 8GB 以上推荐 - 系统: Windows / Linux / macOS #### 安装依赖 pip install -r requirements.txt #### 运行程序 cd 美术资产保护系统 python 美术资产保护系统.py #### 运行效果验证系统 cd 效果验证系统 python 效果验证系统.py ### 使用说明 1. **防护处理**: 切换到「防护工具」标签页，选择图片，选择模式，点击开始防护 2. **效果验证**: 切换到「效果验证」标签页，选择原图和防护图，验证防护效果 3. **盲水印** (V2/V3): 切换到「盲水印」标签页，嵌入或提取水印 4. **ChainGuard** (V3): 切换到「ChainGuard 链智护」标签页，提取或比对图像指纹 ### 目录结构 art-asset-protection-system/ ├── 美术资产保护系统/ # 主程序目录 │ ├── 美术资产保护系统.py # 主程序入口 │ ├── 盲水印处理.py # 盲水印模块 │ ├── ChainGuard 处理.py # ChainGuard 模块 │ ├── MaskSource 介绍.py # MaskSource 模块 │ └── ... ├── 效果验证系统/ # 效果验证系统 │ ├── 效果验证系统.py # 验证系统主程序 │ └── ... ├── requirements.txt # Python 依赖 ├── LICENSE # MIT 许可证 ├── OPEN_SOURCE.md # 开源项目引用声明 ├── .gitignore # Git 忽略配置 └── README.md # 项目说明 ### 许可证 本项目采用 MIT 许可证 - 详见 [LICENSE](LICENSE) 文件。 ## English Description A complete AI protection and copyright protection solution for art assets, providing multi-layer protection algorithms, effect verification, blind watermarking, and image fingerprint comparison. ### Features #### Five-Layer Protection Algorithms - **Fawkes Feature Protection** - Interferes with AI feature extraction, protects image features - **Glaze Style Protection** - Protects artistic style from being mimicked by AI - **Foolbox Adversarial Noise** - Adds adversarial noise to interfere with AI recognition - **OpenStego Steganography** - LSB steganography with misleading data injection - **Blind Watermark** - DWT-DCT-SVD based invisible digital watermark #### Effect Verification System - Feature vector Euclidean distance verification - Artistic style similarity verification - Structural recognition accuracy verification - Steganographic data error rate verification - Visual verification report generation #### Blind Watermark (V2/V3) - Text watermark embedding and extraction - Image watermark embedding and extraction - Password encryption protection #### ChainGuard (V3) - Single image 2048-bit feature fingerprint extraction - Dual image fingerprint similarity comparison - Rapid copyright piracy detection #### MaskSource Model Tracing (V3) - AI model watermark protection introduction - Dual-shield dual-tracing technology system ### Version Comparison | Version | Protection Tools | Verification | Blind Watermark | ChainGuard | MaskSource | |---------|-----------------|--------------|-----------------|------------|------------| | V1 | [x] | [x] | [ ] | [ ] | [ ] | | V2 | [x] | [x] | [x] | [ ] | [ ] | | V3 | [x] | [x] | [x] | [x] | [x] | ### Quick Start #### Requirements - Python 3.8+ - RAM: 8GB+ recommended - OS: Windows / Linux / macOS #### Install Dependencies pip install -r requirements.txt #### Run the Program cd 美术资产保护系统 python 美术资产保护系统.py #### Run Verification System cd 效果验证系统 python 效果验证系统.py ### Usage Guide 1. **Protection**: Go to the "Protection Tools" tab, select an image, choose a mode, and click Start Protection 2. **Verification**: Go to the "Effect Verification" tab, select original and protected images to verify effectiveness 3. **Blind Watermark** (V2/V3): Go to the "Blind Watermark" tab to embed or extract watermarks 4. **ChainGuard** (V3): Go to the "ChainGuard" tab to extract or compare image fingerprints ### Project Structure art-asset-protection-system/ ├── 美术资产保护系统/ # Main program directory │ ├── 美术资产保护系统.py # Main program entry │ ├── 盲水印处理.py # Blind watermark module │ ├── ChainGuard 处理.py # ChainGuard module │ ├── MaskSource 介绍.py # MaskSource module │ └── ... ├── 效果验证系统/ # Effect verification system │ ├── 效果验证系统.py # Verification system main program │ └── ... ├── requirements.txt # Python dependencies ├── LICENSE # MIT License ├── OPEN_SOURCE.md # Open source references ├── .gitignore # Git ignore config └── README.md # Project README ### License This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details. ### Acknowledgments Thanks to the following open source projects: - [blind-watermark](https://github.com/guofei9987/blind_watermark) - Invisible digital watermark - [Pillow](https://github.com/python-pillow/Pillow) - Image processing - [psutil](https://github.com/giampaolo/psutil) - System information detection - [ChainGuard](https://github.com/panda-lsy/gradio-ChainGuard-feature_extractor) - Image feature extraction - [MaskSource](https://github.com/zhanjinxiang/MaskSource) - AI model watermark tracing See [OPEN_SOURCE.md](OPEN_SOURCE.md) for details.

标签：AI对抗样本, Python, 图像指纹, 图像隐写, 数字水印, 数字版权保护, 无后门, 逆向工具