MUMBAI, India, Jan. 2 -- Intellectual Property India has published a patent application (202541123092 A) filed by Dayananda Sagar University, Bengaluru, Karnataka, on Dec. 6, 2025, for 'noise-resilient handwritten language recognition method integrating handcrafted and deep spatial-temporal features.'
Inventor(s) include Benaka Santhosha. S; Ranjima P; Vinitha. V; Dr. Navya. R; Santhosh Kumar R; K. Sudha Deepthi; Shivamma D; Kavyashree I Pattan; Dharmendra D P; and Dr. Santosh Kumar J.
The application for the patent was published on Jan. 2, under issue no. 01/2026.
According to the abstract released by the Intellectual Property India: "The invention relates to a noise-resilient method for handwritten language recognition using a hybrid feature-extraction framework that combines handcrafted descriptors with deep spatial-temporal learning. The method includes pre-processing handwritten input images through noise filtering, binarization, skew correction, and normalization. Handcrafted features comprising Histogram of Oriented Gradients (HOG), Scale-Invariant Feature Transform (SIFT), and Zernike Moments are extracted to capture structural and shape-based properties. In parallel, spatial and temporal features are learned using a Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) architecture. The outputs from the handcrafted and deep-learning pathways are fused to generate a composite feature vector, which is subsequently classified to identify the script or language of the handwritten data. The method provides improved recognition accuracy in the presence of noise, distortions, illumination variations, and inconsistent handwriting styles. The invention is applicable to multilingual handwritten document processing, digitization, and integration with OCR systems."
Disclaimer: Curated by HT Syndication.
