MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202631086335 A) filed by Narula Institute Of Technology on July 14, 2026, for “a Face Spoofing Detection System”.
Inventors include Prianka Dey; Dr. Ratul Chowdhury; Dr. Jayanta Pal; Shubhodeep Singha; Sweety Das; Ritam Dey; and Kankodip Som.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: Abstract A Face Anti-Spoofing Detection System and associated methods are disclosed for securing biometric authentication environments against presentation attacks, such as photographic prints, digital video replays, 3D masks, and deepfakes. The system ingests facial image data or continuous video streams through an optical sensor, executing real-time face detection, spatial landmark alignment, and contrast-limited illumination normalization to isolate a standardized facial region. A dual- branch parallel feature extraction engine processes the facial region using an integrated deep convolutional neural network (ResNet-18) appearance branch and a statistical Local Binary Pattern (LBP) texture branch to simultaneously capture macro-geometric structures and microscopic surface micro-textures. The extracted feature vectors are integrated via adaptive feature fusion layers to yield a unified multi-dimensional descriptor. A regularized classification engine converts the unified descriptor into a liveness probability, which is refined through temporal moving average filtering and majority voting to generate an ultra-reliable authentication verdict.
Disclaimer: Curated by HT Syndication.
