Development of an independent component analysis (ICA) based model for military target detection using hyperspectral data
Implementing Organization
NIT Kurukshetra https://nitkkr.ac.in/
Principal Investigator
Dr. (Col) KC Tiwari
Delhi Technological University, New Delhi
Co-PIs:
Dr. Kaustubh Ranjan Singh
DTU, New Delhi
Dr. Rubina Vohra BVCE, New Delhi
NIT Kurukshetra https://nitkkr.ac.in/
kcchtphd@gmail.com
CO-Principal Investigator
P Satyanaryana, NRSC Hyderabad
NIT Kurukshetra https://nitkkr.ac.in/
Project Overview
This proposal aims to develop a novel super-resolution modeling framework for hyperspectral images that enhances spatial resolution while preserving spectral fidelity, addressing a key limitation of spaceborne hyperspectral sensors. Hyperspectral missions such as HysIS provide rich spectral information for Earth and planetary observations but are constrained by coarse spatial sampling due to sensor, orbital, and telemetry limitations. The proposed research builds upon classical spectral unmixing and super-resolution mapping approaches and advances them through joint spectral–spatial modeling with embedded physical constraints.
The investigation will develop an adaptive super-resolution model that integrates abundance-based spectral unmixing with spatial contextual reconstruction, ensuring conservation of material fractions and minimization of spectral distortion. Where appropriate, learning-based spatial refinement will be incorporated under physics-guided constraints. The model will be validated using hyperspectral data, airborne benchmark datasets, and synthetically degraded scenes. Quantitative evaluation will be carried out using established spectral and spatial metrics, along with application-driven assessments.
Expected outcomes include a mission-ready, computationally efficient super-resolution module suitable for ground processing chains, improved sub-pixel target and anomaly detection, and enhanced thematic products for agriculture, mineral and soil mapping, and water quality assessment. The project is entirely algorithmic, requires no payload development, and can be executed using existing institutional facilities, delivering direct value to current and future hyperspectral missions.
Achievements
The project being undertaken by ISRO in collaboration with academia is directly or indirectly beneficial to the society.