基于特征融合的JPEG图像双压缩检测研究

Digital Image Double Recompression of JPEG Images Based on Feat ure Fusion

在数字图像取证领域,如何高效甄别图像是否被篡改已经成为相关科研工作者的研究重点。笔者介绍了一直双压缩图像的检测方法,从特征融合的角度出发,在已有的篡改检测方法中加入特征融合和特征降维的步骤,并引入机器学习方法,依据统计模型来得到最优的检测率。实验表明,基于双压缩的数字图像取证准确率更高,具有更强的鲁棒性。

In the field of digital image forensics,how to effectively identify whether the image has been tampered has become the research focus of relevant researchers.From the point of view of feature fusion,the steps of feature fusion and feature dimensionality reduction are added to the existing tamper detection methods,and machine learning method is introduced to get the optimal detection rate according to the statistical model.Experiments show that the digital image forensics based on double compression has higher accuracy and stronger robustness.