融合注意力与FCN的图像修复取证  

Images inpainting forensics fused with attention mechanism and FCN

作　　者：沈万里[1] 张玉金 王永琦[1] 胡万 孙冉 郭静 SHEN Wanli;ZHANG Yujin;WANG Yongqi;HU Wan;SUN Ran;GUO Jing(School of Electronic and Electrical Engineering,Shanghai University of Engineering Science,Shanghai 201620,China)

机构地区：[1]上海工程技术大学电子电气工程学院,上海201620

出　　处：《智能计算机与应用》2022年第8期145-147,151,共4页Intelligent Computer and Applications

基　　金：上海市自然科学基金(17ZR1411900)。

摘　　要：图像修复检测是数字图像取证领域中一个非常具有挑战性的问题,为了使图像在被修复后依然保持视觉上的光滑性,常用基于样本块的修复方法来填充被篡改部分。本文提出一种融合注意力机制与全卷积神经网络(Full Convolution Network,FCN)的图像修复取证方法,首先对修复图像进行深度特征提取,再对最终的特征图进行权值分配,以获取最有效的特征图,提高检测效果。为了验证本文方法的可行性,创建了一个修复图像数据库,并通过不同性能指标来评估所提出取证网络的性能。Image inpainting forensics is a challenging problem in the field of digital image forensics.In order to maintain the visual smoothness of the image after inpainting,the inpainting method based on sample blocks is often used to fill the tampered part.This paper proposes an image inpainting forensics method that combines the attention mechanism and Full Convolutional Neural Network(FCN).First,deep feature extraction is performed on the inpainting image,and then the final feature map is weighted to obtain the most effective feature maps and improve the detection effect.In order to verify the feasibility of the method in this paper,a repaired image database is created,and the performance of the proposed forensics network is evaluated through different performance indicators.

关 键 词：数字图像取证 注意力机制 全卷积神经网络 图像修复取证 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]