基于改进的RRU-Net的图像篡改检测算法  

Improved RRU-Net based image tampering detection algorithm

作　　者：卫星星 吴建斌[1] 涂雅蒙 袁林锋 WEI Xingxing;WU Jianbin;TU Yameng;YUAN Linfeng(College of Physical Science and Technology,Central China Normal University,Wuhan 430079,China;Wuhan Maritime Communication Research Institute,China Shipbuilding Industry Group Co.,Ltd.,Wuhan 430079,China)

机构地区：[1]华中师范大学物理科学与技术学院,湖北武汉430079 [2]中国船舶重工集团有限公司武汉船舶通信研究所,湖北武汉430079

出　　处：《微电子学与计算机》2024年第3期53-58,共6页Microelectronics & Computer

基　　金：国家自然科学基金(U1736121);中央高校基本科研业务费专项(CCNU22JC024)。

摘　　要：针对现有深度学习图像篡改检测模型难以利用网络浅层的篡改痕迹特征,导致检测效果差、定位精度低的问题,提出基于改进环形残差U-Net(Ringed Residual U-Net,RRU-Net)的图像篡改检测算法。首先利用分级监督策略设计篡改融合定位模块,将模型分层输出,使深、浅层特征信息充分融合,提高模型对浅层的纹理、边缘信息的敏感性。其次在二元交叉熵基础上对损失函数进行改进,用不同层的损失来衡量总损失。最后在模型中运用组归一化,加快模型收敛速度,同时避免过拟合。在CSAIA和Columbia数据集上的实验结果与RRU-Net相比,F1值分别提高了0.08和0.072,表明该方法具有较高的检测精度,且能有效定位篡改区域。To address the problem that existing deep learning image tampering detection models can hardly make use of the tampering trace features extracted from the shallow layer of the network,resulting in poor detection and low localization accuracy,an image tampering detection algorithm based on the improved Ringed Residual U-Net(RRU-Net)is proposed.Firstly,a hierarchical supervision strategy is used to design a tampering fusion localization module to output the model in layers,so that the deep and shallow feature information is fully fused and the sensitivity of the model to the texture and edge information of the shallow layers is improved.Then,the loss function is improved by adding the parameterβto the binary cross-entropy,and then the total loss is measured by the loss of the different layers.Finally,group normalization is used in the model to speed up the convergence of the network while avoiding overfitting.The experimental results on the CSAIA and Columbia datasets showed an increase in F1 values of 0.08 and 0.072 compared to RRU-Net,respectively.It shows that the algorithm has high detection accuracy and can effectively locate tampered areas.

关 键 词：图像篡改检测 RRU-Net 多层输出 分级监督 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]