Enhanced Image Forgery Detection using a Hybrid Approach: Integration of ELA, CNN, and XGBoost.

Kaur, Sukhmani; Sinha, Nityaa; Jain, Priyasha; Koli, Shruti; Sharma, Arun; Lathwal, Anjali

In light of the escalating prevalence of digital image forgery facilitated by advanced editing tools and widespread sharing on online platforms, the demand for effective forgery detection techniques has surged. This research introduces an approach to digital image forgery detection, employing a multi-stage architecture involving ELA (Error Level Analysis), CNN (Convolutional Neural Networks), and XGBoost. The ELA technique is initially applied to identify tampered areas within an image, followed by CNN for feature extraction. The feature vectors are then fed into an XGBoost classifier, categorizing images as either authentic or forged. This multi-stage process works towards enhancing the detection accuracy and efficiency of forged image detection. The proposed algorithm achieved notable accuracy levels of 90.83%, 96.82%, and 82.82% on the CASIA v1, CASIA v2, and MISD datasets respectively.

CONVOLUTIONAL neural networks; FEATURE extraction; FORGERY; ALGORITHMS

International Journal of Performability Engineering, 2024, Vol 20, Issue 6, p367

0973-1318

Academic Journal

10.23940/ijpe.24.06.p4.367378