Reinforcement learning of non-additive joint steganographic embedding costs with attention mechanism  

作　　者：Weixuan TANG Bin LI Weixiang LI Yuangen WANG Jiwu HUANG 

机构地区：[1]Institute of Artificial Intelligence and Blockchain,Guangzhou University,Guangzhou 510006,China [2]Guangdong Key Laboratory of Intelligent Information Processing,Shenzhen Key Laboratory of Media Security,Shenzhen University,Shenzhen 518060,China [3]Shenzhen Institute of Artificial Intelligence and Robotics for Society,Shenzhen 518060,China [4]School of Computer Science and Cyber Engineering,Guangzhou University,Guangzhou 510006,China

出　　处：《Science China(Information Sciences)》2023年第3期269-282,共14页中国科学（信息科学）（英文版）

基　　金：supported by National Natural Science Foundation of China(Grant Nos.62002075,61872244,61872099,U19B2022);Guangdong Basic and Applied Basic Research Foundation(Grant No.2019B151502001);Shenzhen R&D Program(Grant No.JCYJ20200109105008228).

摘　　要：Image steganography is the art and science of secure communication by concealing information within digital images.In recent years,the techniques of steganographic cost learning have developed rapidly.Although the existing methods can learn satisfactory additive costs,the interplay of different pixels’embedding impacts has not been considered,so the potential of learning may not be fully exploited.To overcome this limitation,in this paper,a reinforcement learning paradigm called JoPoL(joint policy learning)is proposed to extend the idea of additive cost learning to a non-additive situation.JoPoL aims to capture the interactions within pixel blocks by defining embedding policies and evaluating contributions of embedding impacts on a block level rather than a pixel level.Then,a policy network is utilized to learn optimal joint embedding policies for pixel blocks through interactions with the environment.Afterwards,these policies can be converted into joint embedding costs for practical message embedding.The structure of the policy network is designed with an effective attention mechanism and incorporated with the domain knowledge derived from traditional non-additive steganographic methods.The environment is responsible for assigning rewards according to the impacts of the sampled joint embedding actions,which are evaluated by the gradient information of a neural network-based steganalyzer.Experimental results show that the proposed non-additive method JoPoL significantly outperforms the existing additive methods against both feature-based and CNN-based steganalzyers over different payloads.

关 键 词：information hiding non-additive steganography STEGANALYSIS cost learning image processing 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]