MUMBAI, India, Aug. 17 -- Intellectual Property India has published a patent application (202611081714 A) filed by Manipal University Jaipur on July 01, 2026, for Mm-Dec: A Multi-Modal Deep Embedded Clustering System For Mineral Identification In Hyperspectral Satellite Imagery.
Inventors include Priyanka Nair; and Dr. Devesh Kumar Srivastava.
The application for the patent was published on August 14, 2026, under issue no. 33/2026.
Abstract: The present invention relates to a Multi-Modal Deep Embedded Clustering (MM-DEC) system for unsupervised mineral detection from hyperspectral satellite data. The System comprises(100) a data ingestion module (110) configured to acquire hyperspectral image datasets; a preprocessing module (120) configured to improve data quality and extract meaningful spectral information; a multi-modal feature extraction module to process spectral reflectance vectors associated with individual pixels, to process spatial neighborhoods surrounding each pixel, to extracts texture descriptors including contrast, correlation, energy, homogeneity etc.; feature fusion and attention mechanism (160) combines feature extraction vectors into a unified multimodal representation; a deep embedded clustering (dec) module (170) configured to perform unsupervised clustering; a cluster refinement module (CRM) (180) configured to enhance clustering robustness and spatial coherence; and an output generation module (190) configured to generate a mineralogical map indicating spatial distributions of identified mineral clusters. The present system automates mineral prospecting by analyzing hyperspectral satellite imagery to generate digital mineral maps and identify potential mineral targets, supporting efficient mineral exploration, resource assessment, and geological planning.
Disclaimer: Curated by HT Syndication.
