基于改进DeepLabV3+的图像篡改检测技术

Image Tampering Detection Technology Based on Improved DeepLabV3+

现有的图像篡改检测算法在面对未知篡改时,模型的鲁棒性较差,并且篡改区域定位精度较低,小尺度篡改区域检测效果较差,导致模型漏报率较高。针对上述问题,提出了基于改进DeepLabV3+的图像篡改检测模型。模型利用可学习的特征提取器从篡改图像中学习统一的篡改特征;引入注意力模块,强化对篡改特征学习能力,使用ASPP模块提取多尺度特征提高小尺度篡改区域检测率;利用空洞卷积和特征融合模块提高篡改区域定位精度。实验结果表明,提出的方法优于现有的主流方法。

In the face of unknown tampering,the existing image tampering detection algorithm has poor robustness,and the accuracy of tampering area positioning is low,the detection effect of small-scale tampering area is poor,leading to a high failure rate of the model.To solve the above problems,an image tampering detection model based on improved DeepLabV3+is proposed.The model uses a learnable feature extractor to learn unified tamper features from tampered images;introduces an attention module to strengthen the ability to learn tamper features,uses the ASPP module to extract multi-scale features to improve the detection rate of small-scale tampered regions;uses atrous convolution and feature fusion module improve the accuracy of tampering area posi⁃tioning.The experimental results show that the proposed method is better than the existing mainstream methods.