A lightweight image splicing tampering localization method based on MobileNetV2 and SRM. Issue 6 (14th February 2023)

• Record Type: Journal Article
• Authors: Shi, Xiaoqian; Li, Ping; Wu, Hao; Chen, Qidong; Zhu, Haoyu
• Is Part Of: IET image processing. Volume 17:Issue 6(2023)
• Page Start: 1883
• Page End: 1892
• Publication Date: 2023-02-14
• Subjects: dual‐stream network -- image tampering localization -- lightweight convolutional network; Image processing -- Periodicals; 621.36705
• DOI: 10.1049/ipr2.12763
• Languages: English
• ISSNs: 1751-9659

Abstract

Abstract: The architectures of many state‐of‐the‐art local tempering detection models are complexity, and the training process of those models is also time‐consuming. Therefore, this paper constructs a lightweight local tampering detection method based on the convolutional network MobileNetV2 and a dual‐stream network. Specifically, the algorithm first improves the MobileNetv2, which not only reduces the multiple of its downsampling operator to retain richer traces of image tampering, but also introduces the dilated convolution in it to expand the receptive field of feature maps. The dual‐stream network uses RGB stream to extract image tampering features such as strong contrast difference and unnatural tampered boundaries, and implements spatial rich model (SRM) stream to extract image tampered area and noise features of real area. Finally, the features extracted from two streams are fused through an improved attention mechanism called parallel convolutional block attention module (CBAM), which can improve the sensitivity of the model to important features in RGB and SRM. The experimental results show that the proposed algorithm still has higher positioning accuracy than some existing algorithms, while achieving lightweight. Abstract : First, the algorithm improves the lightweight convolutional network MobileNetv2 not only by reducing the multiple of its downsampling operator so as to retain richer traces of image tampering, but also introduces the dilated convolution in it to expand the receptive field of feature maps. In addition, a two‐stream network is constructed in the model, which uses RGB stream to extract image tampering features such as strong contrast difference and unnatural tampered boundaries and uses SRM stream to extract image tampered area and noise features of real area. Finally, the extracted two‐stream features are fused through an improved attention mechanism.