文献类型：期刊文献

英文题名：Covert Communication in Sanitized Online Social Networks

作者：Zhu, Zhiying[1,2]; Ba, Zhongjie[3]; Li, Guobiao[4]; Qian, Zhenxing[4]; Zhang, Xinpeng[4]

机构：[1] East China University of Science and Technology, School of Information Science and Engineering, China; [2] Zhejiang University, State Key Laboratory of Blockchain and Data Security, Hangzhou, China; [3] Zhejiang University, The State Key Laboratory of Blockchain and Data Security, Hangzhou, China; [4] Fudan University, School of Computer Science, China

年份：2025

起止页码：144

外文期刊名：IEEE International Workshop on Information Forensics and Security, WIFS

收录：EI(收录号：20261920658857)

语种：英文

外文关键词：Image processing - Iterative methods - Online systems

摘要：Robust steganography has made significant progress over the years, and it is now highly convenient to transmit secret messages through lossy channels such as Online Social Networks (OSNs). As a countermeasure, recent studies propose sanitizing networks to indiscriminately process uploaded images on OSNs, which are effective in interrupting illegal covert communications. Unfortunately, these sanitizing networks also corrupt the secret data of certain legitimate and justified needs. To this end, we present in this paper a novel scheme tailored for authorized users to conduct covert communication in OSNs deployed with sanitizing networks. Specifically, we first joined the sanitizing network with a decoder whose weights are generated according to a secret seed (i.e., key). Then, we iteratively optimize a cover image using the jointed network until the updated cover image (i.e., stego image) triggers the jointed network to generate the specific output corresponding to the secret. As such, only authorized receivers who possess the seed could rebuild the decoder to recover the secret data from the sanitized stego images. Various experiments have been conducted to demonstrate the advantage of our proposed method for covert communication in sanitized OSNs. ? 2025 IEEE.