Classification of complex environments using pixel level fusion of satellite data
- PDF / 3,939,159 Bytes
- 33 Pages / 439.37 x 666.142 pts Page_size
- 91 Downloads / 197 Views
Classification of complex environments using pixel level fusion of satellite data Amol D. Vibhute 1 & Karbhari V. Kale 2 & Sandeep V. Gaikwad 2 & Rajesh K. Dhumal 1 & Ajay D. Nagne 3 & Amarsinh B. Varpe 2 & Dhananjay B. Nalawade 2 & Suresh C. Mehrotra 2 Received: 13 May 2019 / Revised: 13 April 2020 / Accepted: 22 April 2020 # Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract
The present study reports classification and analysis of composite land features using fusion images obtained by fusing two original hyperspectral and multispectral datasets. The high spatial-spectral resolution, multi-instrument and multi-period satellite images were used for fusion. Three pixel level fusion based techniques, Color Normalized Spectral Sharpening (CNSS), Principal Component Spectral Sharpening Transform (PCSST) and Gram-Schmidt Transform (GST), were implemented on the datasets. Performance evaluations of three fusion algorithms were done using classification results. The Support Vector Machine (SVM) and Gaussian Maximum Likelihood Classification (MLC) were used for classification using five types of images, viz. hyperspectral, multispectral and three fused images. Number of classes considered was eight. Sufficient number of ground field data for each class has also been acquired which was needed for supervise based classification. The accuracy was improved from 74.44 to 97.65% when the fused images were considered with SVM classifier. Similarly, the results were improved from 69.25 to 94.61% with original and fused data using MLC classifier. The fusion image technique was found to be superior to the single original image and the SVM is better than the MLC method. Keywords Pixel level fusion . Color normalized spectral sharpening (CNSS) . Dimensionality reduction . Minimum noise fraction . Maximum likelihood classifier . Supervised classification
1 Introduction Presently diverse sensors have been used for earth observation to identify and classify various patterns on the earth surface. Each sensor has its own spatial and spectral attributes in hyperspectral, multispectral and panchromatic (PAN) images [11], elevation in Light
* Amol D. Vibhute [email protected] Extended author information available on the last page of the article
Multimedia Tools and Applications
Detection And Ranging (LiDAR) data, amplitude and phase in Synthetic Aperture Radar (SAR) systems. However, analysis of satellite images are still challenging task due to assorted effect of various objects [9]. The heterogeneous and composite terrestrial patterns makes difficult for single source remotely sensed images to meet all requirements of land classification. Nevertheless, detection and classification of heterogeneous-assorted terrestrial patterns is useful application for urban land use/cover classification [28, 30], classification of complex forest areas [8], multiple crop classification [10], etc. The single source technology and low spatial-spectral data do not provide the details of heterogeneous and assorted earth
Data Loading...
