DeepNet: Protection of deepfake images with aid of deep learning networks

IF 4.2 3区计算机科学 Q2 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Image and Vision Computing Pub Date : 2025-04-02 DOI:10.1016/j.imavis.2025.105540

Divyanshu Awasthi , Priyank Khare , Vinay Kumar Srivastava , Amit Kumar Singh , Brij B. Gupta

{"title":"DeepNet: Protection of deepfake images with aid of deep learning networks","authors":"Divyanshu Awasthi , Priyank Khare , Vinay Kumar Srivastava , Amit Kumar Singh , Brij B. Gupta","doi":"10.1016/j.imavis.2025.105540","DOIUrl":null,"url":null,"abstract":"<div><div>In the present information age, multimedia security has become a challenging task. Especially increased usage of images as multimedia data has been a key aspect in this digital transmission era. Deep fake detection of images is a real-time problem which needs to be focused. To resolve this challenge, a novel deep fake detection algorithm is proposed in this article. The presented research uses the Viola-Jones detection algorithm for efficient deep fake image detection. To protect the integrity of these images, the multiresolution domain approach is effectively utilized with redundant discrete wavelet transform (RDWT) and multiresolution singular value decomposition (MSVD). Discrete cosine transform (DCT) is applied for the extraction of frequency components. An adaptive neuro-fuzzy inference system (ANFIS)-based optimization is applied to attain the optimum weighing factor (WF). This WF exhibits a better trade-off among attributes of watermarking. Furthermore, authentication is successfully implemented with the aid of various deep learning models such as SqueezeNet, EfficientNet-B0, ResNet-50 and InceptionV3. This implementation explores the various aspects related to the ownership assertion. Analysis of comprehensive simulation results depicts the effectiveness of the proposed technique over different prevailing techniques. With the development of the proposed technique, deep fake image detection can easily be realized and safeguards the images. The average percentage improvement in the imperceptibility of the proposed technique is 52.14% and for robustness is 7.51%.</div></div>","PeriodicalId":50374,"journal":{"name":"Image and Vision Computing","volume":"158 ","pages":"Article 105540"},"PeriodicalIF":4.2000,"publicationDate":"2025-04-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Image and Vision Computing","FirstCategoryId":"94","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0262885625001283","RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE","Score":null,"Total":0}

引用次数: 0

Abstract

In the present information age, multimedia security has become a challenging task. Especially increased usage of images as multimedia data has been a key aspect in this digital transmission era. Deep fake detection of images is a real-time problem which needs to be focused. To resolve this challenge, a novel deep fake detection algorithm is proposed in this article. The presented research uses the Viola-Jones detection algorithm for efficient deep fake image detection. To protect the integrity of these images, the multiresolution domain approach is effectively utilized with redundant discrete wavelet transform (RDWT) and multiresolution singular value decomposition (MSVD). Discrete cosine transform (DCT) is applied for the extraction of frequency components. An adaptive neuro-fuzzy inference system (ANFIS)-based optimization is applied to attain the optimum weighing factor (WF). This WF exhibits a better trade-off among attributes of watermarking. Furthermore, authentication is successfully implemented with the aid of various deep learning models such as SqueezeNet, EfficientNet-B0, ResNet-50 and InceptionV3. This implementation explores the various aspects related to the ownership assertion. Analysis of comprehensive simulation results depicts the effectiveness of the proposed technique over different prevailing techniques. With the development of the proposed technique, deep fake image detection can easily be realized and safeguards the images. The average percentage improvement in the imperceptibility of the proposed technique is 52.14% and for robustness is 7.51%.

查看原文本刊更多论文

DeepNet：借助深度学习网络保护deepfake图像

在当今信息时代，多媒体安全已成为一项具有挑战性的任务。特别是越来越多的图像作为多媒体数据的使用已经成为这个数字传输时代的一个关键方面。图像的深度假检测是一个需要关注的实时问题。为了解决这一问题，本文提出了一种新的深度假检测算法。本研究使用Viola-Jones检测算法进行高效的深度假图像检测。为了保护图像的完整性，将多分辨率域方法与冗余离散小波变换（RDWT）和多分辨率奇异值分解（MSVD）有效地结合起来。采用离散余弦变换（DCT）提取频率分量。采用基于自适应神经模糊推理系统（ANFIS）的优化方法获得最优权重因子（WF）。该WF在水印属性之间表现出更好的权衡。此外，通过多种深度学习模型（如SqueezeNet、EfficientNet-B0、ResNet-50和InceptionV3）成功实现了身份验证。这个实现探索了与所有权断言相关的各个方面。综合仿真结果的分析表明了该技术相对于其他主流技术的有效性。随着该技术的发展，可以很容易地实现深度假图像检测并保护图像。所提出的技术在不可感知性方面的平均改进百分比为52.14%，鲁棒性方面的平均改进百分比为7.51%。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Image and Vision Computing 工程技术-工程：电子与电气

CiteScore

8.50

自引率

8.50%

发文量

143

审稿时长

7.8 months

期刊介绍： Image and Vision Computing has as a primary aim the provision of an effective medium of interchange for the results of high quality theoretical and applied research fundamental to all aspects of image interpretation and computer vision. The journal publishes work that proposes new image interpretation and computer vision methodology or addresses the application of such methods to real world scenes. It seeks to strengthen a deeper understanding in the discipline by encouraging the quantitative comparison and performance evaluation of the proposed methodology. The coverage includes: image interpretation, scene modelling, object recognition and tracking, shape analysis, monitoring and surveillance, active vision and robotic systems, SLAM, biologically-inspired computer vision, motion analysis, stereo vision, document image understanding, character and handwritten text recognition, face and gesture recognition, biometrics, vision-based human-computer interaction, human activity and behavior understanding, data fusion from multiple sensor inputs, image databases.