海洋测绘Issue(4)：55-60,6.DOI:10.3969/j.issn.1671-3044.2025.04.011
一种基于残差神经网络的伪造遥感影像判别方法
A residual-neural-network-based discrimination method for fake remote sensing
摘要
Abstract
Considering the security risk of fake remote sensing images influencing on the open source geographic information on Internet,this paper describes the fake remote sensing image clearly,systematically reviews the methods of creating fake remote sensing images,summarizes common types of fake remote sensing images,and designs a discrimination method of fake remote sensing images based on the residual neural networks.The proposed method can learn features from the remote sensing images and fake remote sensing images,and distinguish the remote sensing images,fake remote sensing images and the types of fake remote sensing images automatically,accurately and effectively.Experiments are designed based on the open source remote sensing datasets and existing algorithms of creating fake remote sensing images.The experimental results show that:the proposed method can accurately distinguish remote sensing images from fake remote sensing images and identify the types of fake remote sensing images.Additionally,the designed residual neural network discrimination model is better than some other popularly used neural-network-based discrimination models,and is more suitable for the application of discriminating fake remote sensing images.The effectiveness and superiority of the proposed method have been verified.This research is able to support the quality assessment of remote sensing image,the security risk assessment geographic information and some other related fields.关键词
地理信息安全/伪造遥感影像/识别真伪影像/生成式人工智能/残差神经网络Key words
geographic information security/created fake remote sensing image/discrimination of real and fake remote sensing images/generative artificial intelligence/residual neural networks分类
天文与地球科学引用本文复制引用
陈庆基,杜佳威,杨世环,张珂..一种基于残差神经网络的伪造遥感影像判别方法[J].海洋测绘,2025,(4):55-60,6.基金项目
国家自然科学基金项目(42301504). （42301504）
