1 July 2007 Application of principal component analysis and information fusion technique to detect hotspots in NOAA/AVHRR images of Jharia coalfield, India
Author Affiliations +
J. of Applied Remote Sensing, 1(1), 013523 (2007). doi:10.1117/1.2771256
Abstract
Present paper proposes an algorithm for hotspot (sub-surface fire) detection in NOAA/AVHRR images in Jharia region of India by employing Principal Component Analysis (PCA) and fusion technique. Proposed technique is very simple to implement and is more adaptive in comparison to thresholding, multi-thresholding and contextual algorithms. The algorithm takes into account the information of AVHRR channels 1, 2, 3, 4 and vegetation indices NDVI and MSAVI for the required purpose. Proposed technique consists of three steps: (1) detection and removal of cloud and water pixels from preprocessed AVHRR image and screening out the noise of channel 3, (2) application of PCA on multichannel information along with vegetation index information of NOAA/AVHRR image to obtain principal components, and (3) fusion of information obtained from principal component 1 and 2 to classify image pixels as hotspots. Image processing techniques are applied to fuse information in first two principal component images and no absolute threshold is incorporated to specify whether particular pixel belongs to hotspot class or not, hence, proposed method is adaptive in nature and works successfully for most of the AVHRR images with average 87.27% detection accuracy and 0.201% false alarm rate while comparing with ground truth points in Jharia region of India.
Rohit Singh Gautam, Dharmendra Singh, Ankush Mittal, "Application of principal component analysis and information fusion technique to detect hotspots in NOAA/AVHRR images of Jharia coalfield, India," Journal of Applied Remote Sensing 1(1), 013523 (1 July 2007). http://dx.doi.org/10.1117/1.2771256
JOURNAL ARTICLE
17 PAGES


SHARE
KEYWORDS
Image fusion

Principal component analysis

Clouds

Satellites

Vegetation

Image processing

Information fusion

Back to Top