DSpace Arşivi :: by Yazar "Ata, Oǧuz" değerine göre listeleniyor

Listeleniyor 1 - 1 / 1

Forensic Image Enhancement and Forgery Detection Using Advanced Image Processing Techniques and Convolutional Neural Networks
(Institute of Electrical and Electronics Engineers Inc., 2025) Nofan, Mohammed Waleed; Ata, Oǧuz
Nowadays, there is a wide use of modern techniques and technology in general, which constitutes an important innovation. Manipulating, changing and modifying digital images has become very popular and relatively easy, which has created a great challenge and concern, especially in cases that rely on digital images as the main evidence for issuing judgments. In order to improve the quality of images and accurately identify fraud cases, a talented system has been invented to conduct forensic investigations by combining modern image processing techniques and Convolutional Neural Networks (CNN). In this experiment, Google Colab was used as an experimental platform to apply a pixel-based approach to extract features from three separate experiments. An approach to forgery detection using pixel base algorithm was tested in relation to multiple conditions within separate experiments involving CNNs integration. Experiment1, which had a rudimentary CNN architecture, achieved an accuracy of 94.53% and validation accuracy of 94.54%. Experiment 2, with a sophisticated CNN structure, resulted in an accuracy of 94.91%, and a validation accuracy of 94.92%.