Scholarlink

<p>As artificial intelligence and multimedia technologies advance, their potential for misuse also increases, with image forgery-particularly through techniques like copy-move-emerging as a significant concern. While copy-move forgeries can produce high-quality images, they also pose serious security risks, especially with the widespread dissemination of forged content on social media. Detecting and identifying these forgeries is a challenging task. This article proposes a feature fusion-based optimization algorithm for detecting copy-move image forgery, utilizing deep learning through a transfer learning approach. A Convolutional Neural Network (CNN) is employed for feature extraction. The algorithm is evaluated on the MICC-F220, MICC-F2000, CoMoFoD, and CASIA v2 datasets, with performance assessed using accuracy, precision, recall, F1-score, and specificity. Experimental results demonstrate that the proposed method outperforms existing deep learning-based forgery detection algorithms, achieving improvements of 2–4% across all evaluation metrics.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Modelling of features of fusion using a hybrid swarm optimization algorithm with deep learning methodology for copy-move image forgery detection

  • 期刊论文

  • 发表日期：

    2025年9月15日

查看全文
  • Alesh Kumar Sharma,
  • Ritu Tiwari,
  • Rahul Dixit,
  • Mahendra Pratap Yadav

摘要

As artificial intelligence and multimedia technologies advance, their potential for misuse also increases, with image forgery-particularly through techniques like copy-move-emerging as a significant concern. While copy-move forgeries can produce high-quality images, they also pose serious security risks, especially with the widespread dissemination of forged content on social media. Detecting and identifying these forgeries is a challenging task. This article proposes a feature fusion-based optimization algorithm for detecting copy-move image forgery, utilizing deep learning through a transfer learning approach. A Convolutional Neural Network (CNN) is employed for feature extraction. The algorithm is evaluated on the MICC-F220, MICC-F2000, CoMoFoD, and CASIA v2 datasets, with performance assessed using accuracy, precision, recall, F1-score, and specificity. Experimental results demonstrate that the proposed method outperforms existing deep learning-based forgery detection algorithms, achieving improvements of 2–4% across all evaluation metrics.