Framework for Localization of Forgery Regions in Image


Abstract views: 31 / PDF downloads: 132

Authors

  • Canberk Şahin Bursa Technical University
  • Mustafa Özden Bursa Technical University

DOI:

https://doi.org/10.59287/icaens.1131

Keywords:

Image forgery, Convolutional Neural Network (CNN), Discrete Cosine Transform (DCT), Discrete Wavelet Transform (DWT), Long-Short Term Memory (LSTM), Intersection Over Union (IOU)

Abstract

With the development of computer technologies, manipulations are made on digital images without leaving a clear trace thanks to image processing software. There is a great need for applications to detect forgered images made with malicious intent in many fields such as politics, law, medicine and military. Many studies have been carried out and various algorithms have been developed to detect forgered regions by detecting forgered images. Today, superior methods are developed by combining traditional image forgery techniques with deep learning techniques. In this study, the Discrete Cosine Transform (DCT) and Discrete Wavelet Transform (DWT) were used together with Convolutional Neural Networks (CNN) to locate the forgered regions of the forgered images. Three different methods were made to locate the forgered region. In the first method, DWT and CNN were used together. In the second method, DCT and CNN were used together. In the last method, DCT and DWT were combined in parallel and used together with CNN.

Author Biographies

Canberk Şahin, Bursa Technical University

Department of Electrical and Electronics Engineering,  Bursa, 16310, Turkey

Mustafa Özden, Bursa Technical University

Department of Electrical and Electronics Engineering,  Bursa, 16310, Turkey

Downloads

Published

2023-07-22

How to Cite

Şahin, C., & Özden, M. (2023). Framework for Localization of Forgery Regions in Image. International Conference on Applied Engineering and Natural Sciences, 1(1), 1071–1078. https://doi.org/10.59287/icaens.1131