MUMBAI, India, Aug. 17 -- Intellectual Property India has published a patent application (202641096404 A) filed by Velammal Institute Of Technology; Mr. Anbumani A; Ms. Pratheeba R S; Mr. R. Venkadesh; Barath R; Mithun Raj T; and Nithish N on August 10, 2026, for Automated Building Extraction From Satellite Images Using Lstm.
Inventors include Mr. Anbumani A; Ms. Pratheeba R S; Mr. R. Venkadesh; Barath R; Mithun Raj T; and Nithish N.
The application for the patent was published on August 14, 2026, under issue no. 33/2026.
Abstract: This project presents a deep learning-based framework for automated building detection from high-resolution satellite images using an LSTM-based architecture. The system begins with the acquisition of satellite imagery from publicly available datasets, followed by pre processing techniques such as noise reduction, contrast enhancement, normalization, and resizing to improve data quality. Feature extraction methods, including edge detection and morphological operations, are applied to capture spatial characteristics of buildings. These extracted features are then converted into sequences and processed by an LSTM network to learn complex spatial dependencies and accurately classify building and non-building regions. A post-processing stage refines the segmentation results using thresholding and morphological filtering to eliminate noise and enhance structural boundaries. Finally, the detected buildings are visualized and overlaid on the original images for clear interpretation. The proposed approach improves detection accuracy and provides an efficient solution for urban planning and remote sensing applications.
Disclaimer: Curated by HT Syndication.
