logo
banner

Journals & Publications

Journals Publications Papers

Papers

An Approach for Hyperspectral Image Classification by Optimizing SVM Using Self Organizing Map
Jul 11, 2018Author:
PrintText Size A A

Title: An Approach for Hyperspectral Image Classification by Optimizing SVM Using Self Organizing Map

Authors: Jain, DK; Dubey, SB; Choubey, RK; Sinhal, A; Arjaria, SK; Jain, A; Wang, HX

Author Full Names: Jain, Deepak Kumar; Dubey, Surendra Bilouhan; Choubey, Rishin Kumar; Sinhal, Amit; Arjaria, Siddharth Kumar; Jain, Amar; Wang, Haoxiang

Source: JOURNAL OF COMPUTATIONAL SCIENCE, 25 252-259; 10.1016/j.jocs.2017.07.016 MAR 2018

Language: English

Abstract: In this paper, an efficient technique for the classification of Hyper-Spectral Images taken form satellite is actualized. The Proposed Methodology is based on the concept of optimizing Support Vector Machine (SVM) using Self Organizing Maps (SOM) and then classification of Interior and Exterior Pixels can be done by the comparing the Posterior Probability of each of the pixel intensities. The Methodology applied here works in two phases, first is to train the important features from the image by Optimizing Support Vector Machine using Self Organizing Maps and second is to find the Interior and Exterior Pixels and Comparing Optimal Threshold and Probability. The Experimental results are performed on two datasets which consists of 16 and 9 classes such as corn-no till, corn, soybeans-no till, corn-min till, soybeans-clean till, soybeans min till alfalfa, grass/trees, grass/pasture, grass/pasture-mowed, oats, hay windrowed wheat, woods, stone-steel towers and building-grass-trees-drives, The proposed Methodology Outperforms better in comparison with the existing Classification methodology in terms of Accuracy and Kappa and Confusion Matrix. (C) 2017 Elsevier B.V. All rights reserved.

ISSN: 1877-7503

IDS Number: GF4LB

Unique ID: WOS:000431933300025

*Click Here to View Full Record