2026-10-14T18:09:36Z https://www.scidb.cn/oai

10.57760/sciencedb.j00173.00044 2026-05-14T09:39:43Z

2026-05-14 A Multi-View Feature Extraction and Dual-Edge Contrastive Learning Approach for Image Forgery Detection-Implementation Code doi:10.57760/sciencedb.j00173.00044 en Solemnly declare: If you use this open source content in papers, books, academic reports and other works, please quote the following documents (the original link has the latest citation format): XU Zhuang, YE Ziyi, PAN Enkang, LIU Chunxiao. A Multi-View Feature Extraction and Dual-Edge Contrastive Learning Approach for Image Forgery Detection[J]. Journal of Electronics & Information Technology, in press. doi: 10.11999/JEIT251271 Authors: XU Zhuang, YE Ziyi, PAN Enkang, LIU ChunxiaoAuthor：School of Computer Science and Technology, Zhejiang Gongshang UniversityDOI：10.11999/JEIT251271Original：https://jeit.ac.cn/cn/article/doi/10.11999/JEIT251271Correspondents: LIU Chunxiao，cxliu@mail.zjgsu.edu.cnOpen source date: April 24(th), 2026Funds: National Natural Science Foundation of China (61976188), Zhejiang Provincial Natural Science Foundation of China (LY24F020004), National College Students Innovation and Entrepreneurship Training Program (202510353027), Zhejiang Provincial College Students Innovation and Entrepreneurship Training Program (S202510353076) Open source content1 A Multi-View Feature Extraction and Dual-Edge Contrastive Learning Approach for Image Forgery Detection-Implementation CodeAbstract:   Objective  Image forgery detection technology holds significant application value in fields such as news auditing and judicial forensic authentication. Aiming at the label conflict problem of existing methods under the definition of pixel-wise classification task, as well as the limitation that tampering cue mining is mostly concentrated in the spatial domain while features from other perspectives are neglected, this paper proposes an improved image forgery detection model based on intra-image inconsistency, and a corresponding image forgery detection algorithm based on multi-view feature extraction and dual-edge contrastive learning. The proposed algorithm is implemented following the design rationale of the aforementioned model. Targeting the inherent limitations of existing approaches, the proposed model can effectively avoid the label conflict problem, enhance the mining capability of tampering cues, improve generalization performance, and overcome the drawbacks of existing methods. Experiments on challenging benchmarks demonstrate that compared with existing mainstream methods, the proposed method achieves an average improvement of 26.0% and 10.1% on the pF1 (per-pixel F1-score) and pIoU (per-pixel Intersection over Union) metrics, respectively. Image Forgery Detection; Contrastive Learning; Multi-View Feature Extraction; Edge Refinement xu zhuang Liu Chunxiao PUBLIC https://api.github.com/licenses/mit dataset Science Data Bank