MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202621093436 A) filed by Parul University Parul Institute Of Engineering Technology; Sardhara Aarti Sureshbhai; and Dr. Vipul Vekariya on August 01, 2026, for An Artifact-Aware Hybrid Deep Learning System And Method For Digital Image Forgery Detection.
Inventors include Sardhara Aarti Sureshbhai; and Dr. Vipul Vekariya.
The application for the patent was published on September 18, 2026, under issue no. 38/2026.
Abstract: The present invention relates to an artifact-aware hybrid deep learning system and method for digital image forgery detection. The invention comprises an Image Acquisition Module for receiving digital images, an Artifact-Aware Preprocessing Module configured to perform Error Level Analysis (ELA), Bayesian denoising, Peak Signal-to-Noise Ratio (PSNR) based quality validation and edge enhancement, a Spatial Feature Extraction Module employing a Convolutional Neural Network (CNN), a Contextual Sequence Learning Module employing a Long Short-Term Memory (LSTM) network, a Feature Fusion Module, a Forgery Classification Module, a Decision Optimization Module, a Performance Monitoring Module and a Forensic Reporting Module. The system integrates artifact-aware preprocessing with hybrid spatial and contextual feature learning to accurately distinguish authentic images from manipulated images. The disclosed invention improves detection accuracy, robustness against image compression and diverse manipulation techniques, cross-dataset generalization and computational efficiency, thereby enabling reliable digital image forgery detection for forensic, security, surveillance, media authentication and judicial applications.
Disclaimer: Curated by HT Syndication.
