Evaluation of Perceptual Hashing Algorithms for Digital Image Authenticity
Mohamed Ben Aissi, Soulaiman El Hazzat, Majid Ben Yakhlef · 2024
The authenticity and integrity of digital images have become paramount in an era of advanced editing tools and widespread manipulation. To address the challenges of detecting tampered images, this research provides a thorough examination of six perceptual hashing techniques. A-Hash, DHash, P-Hash, W-Hash, SVD-Hash, PDQ-Hash. Unlike existing studies that primarily focus on isolated methods, this work provides a comparative analysis to determine their effectiveness in identifying complex tampering approaches such as copy-move and splicing manipulations, using the CASIA V2 dataset.The original contribution of this study lies in its systematic assessment of these algorithms based on three key criteria: robustness to perceptual variations, capacity for discrimination, and authentication ability, measured using the normalized Hamming distance metric.This work not only offers valuable insights into the strengths and weaknesses of perceptual hashing techniques but also serves as a foundational reference for advancing digital forensics and enhancing the reliability of image authentication systems.