科技创新与应用2025，Vol.15Issue(30)：134-137,4.DOI:10.19981/j.CN23-1581/G3.2025.30.032
基于扩散模型的轮廓生成式信息隐藏方法
摘要
Abstract
Aiming at the problem of poor quality of secret images in generative information hiding methods,this paper proposes a contour generative information hiding method based on diffusion model.The sender first extracts contour points from the image,builds a contour candidate list based on the contour points,and then constructs a Huffman code,selects corresponding contour points according to the secret information,uses a differentiable raster to generate a series of contour lines from all contour points,and renders the contour lines to generate a secret contour image.In order to ensure that the dense contour image can well retain the contour features of the original image,CLIP model is used to optimize it.Then,the density carrying contour image is generated by combining a diffusion model without classification guidance.After the receiver obtains the secret image,it first extracts the secret image,then extracts the contour points from it and constructs a Huffman code to extract the secret information.Experimental results show that the overall performance of this method is better than the current mainstream generative image hiding schemes,especially the generated higher-quality cryptography-carrying images,and the extraction accuracy and robustness can be taken into account.关键词
生成式信息隐藏/扩散模型/CLIP模型/载密图像/隐藏方法Key words
generative information hiding/diffusion model/CLIP model/secret image/hiding method分类
计算机与自动化引用本文复制引用
李爽,张春玉,冯宇舒..基于扩散模型的轮廓生成式信息隐藏方法[J].科技创新与应用,2025,15(30):134-137,4.基金项目
2022年国家自然科学基金项目(62262062) （62262062）
