Detecting Burnscar from Hyperspectral Imagery via Sparse Representation with Low-Rank Interference

TL;DR

This paper introduces a novel burnscar detection method for hyperspectral images that uses low-rank and sparse decomposition to effectively separate cloud interference and identify burn areas.

Contribution

The paper presents a new two-step model combining RPCA and low-rank analysis for improved burnscar detection in hyperspectral imagery.

Findings

Effective cloud suppression using RPCA

Accurate burnscar detection demonstrated on MODIS dataset

Enhanced hyperspectral analysis through low-rank representation

Abstract

In this paper, we propose a burnscar detection model for hyperspectral imaging (HSI) data. The proposed model contains two-processing steps in which the first step separate and then suppress the cloud information presenting in the data set using an RPCA algorithm and the second step detect the burnscar area in the low-rank component output of the first step. Experiments are conducted on the public MODIS dataset available at NASA official website.