MUMBAI, India, Aug. 24 -- Intellectual Property India has published a patent application (202631097981 A) filed by National Institute Of Technology Jamshedpur on August 13, 2026, for System And Method For Robust Copy–move Forgery Detection And Pixel-Level Localization In Digital Images.
Inventors include Mr. Shesh Narayan Sahu; Dr. Jitesh Pradhan; and Dr. Dinesh Kumar.
The application for the patent was published on August 21, 2026, under issue no. 34/2026.
Abstract: The system for robust copy–move forgery detection and pixel-level localization in digital images comprises an input image preprocessing unit (102) to normalize pixel intensities and perform a stationary wavelet transform to generate multi-scale frequency detail subbands that are concatenated into a frequency tensor. A dual-branch spatial-frequency feature processing network (104) extracts spatial image representations and high-frequency manipulation artifacts, which are spatially aligned and fused. A contextual relation feature generation unit (108) transforms the fused representations into spatial tokens and computes global contextual relationships through scaled dot-product attention. A copy–move relation processing unit (110) generates an affinity matrix using pairwise cosine similarity, derives a copy-likelihood map, and appends the map to the contextual representation. A central processing unit performs pixel-level forgery localization through progressive upsampling with skip connections and simultaneously executes image-level authenticity classification using global feature pooling and a multi-layer feed-forward network, enabling accurate forgery identification and localization.
Disclaimer: Curated by HT Syndication.
