World Library  


Add to Book Shelf
Flag as Inappropriate
Email this Book

Automatic Generation of Training Data for Hyperspectral Image Classification Using Support Vector MacHine : Volume Xl-7/W3, Issue 1 (29/04/2015)

By Abbasi, B.

Click here to view

Book Id: WPLBN0004015977
Format Type: PDF Article :
File Size: Pages 6
Reproduction Date: 2015

Title: Automatic Generation of Training Data for Hyperspectral Image Classification Using Support Vector MacHine : Volume Xl-7/W3, Issue 1 (29/04/2015)  
Author: Abbasi, B.
Volume: Vol. XL-7/W3, Issue 1
Language: English
Subject: Science, Isprs, International
Collections: Periodicals: Journal and Magazine Collection, Copernicus Publications
Historic
Publication Date:
2015
Publisher: Copernicus Publications, Göttingen, Germany
Member Page: Copernicus Publications

Citation

APA MLA Chicago

Bigdeli, B., Arefi, H., Roessner, S., & Abbasi, B. (2015). Automatic Generation of Training Data for Hyperspectral Image Classification Using Support Vector MacHine : Volume Xl-7/W3, Issue 1 (29/04/2015). Retrieved from http://www.ebooklibrary.org/


Description
Description: Department of Geomatics and Surveying Eng., University of Tehran, Tehran, Iran. An image classification method based on Support Vector Machine (SVM) is proposed on hyperspectral and 3K DSM data. To obtain training data we applied an automatic method relating to four classes namely; building, grass, tree, and ground pixels. First, some initial segments regarding to building, tree, grass, and ground pixels are produced using different feature descriptors. The feature descriptors are generated using optical (hyperspectral) as well as range (3K DSM) images. The initial building regions are created using DSM segmentation. Fusion of NDVI and elevation information assist us to provide initial segments regarding to the grass and tree areas. Also, we created initial segment regarding to ground pixel after geodesic based filtering of DSM and elimination of the non-ground pixels. To improve classification accuracy, the hyperspectral image and 3K DSM were utilized simultaneously to perform image classification. For obtaining testing data, labelled pixels was divide into two parts: test and training. Experimental result shows a final classification accuracy of about 90% using Support Vector Machine. In the process of satellite image classification; provided by 3K camera. Both datasets correspond to Munich area in Germany.

Summary
Automatic Generation Of Training Data For Hyperspectral Image Classification Using Support Vector Machine

 

Click To View

Additional Books


  • Evaluating Sensor Linearity of Chosen In... (by )
  • Comprehensive Evaluation and Analysis of... (by )
  • Epipolar Rectification for Cartosat-1 St... (by )
  • Real Time Integration of Field Data Into... (by )
  • A Multi-sensor Approach to Semi-global M... (by )
  • Ancient Stone Tidal Weirs in Penghu Arch... (by )
  • Uav Linear Photogrammetry : Volume Xl-3/... (by )
  • Investigation on Automatic Change Detect... (by )
  • Housediff: a Map-based Building Change D... (by )
  • The Development of Geospatial Education ... (by )
  • Photogrammetry and Laser Scanning for th... (by )
  • Validation of Spectral Continuity Betwee... (by )
Scroll Left
Scroll Right

 



Copyright © World Library Foundation. All rights reserved. eBooks from World eBook Library are sponsored by the World Library Foundation,
a 501c(4) Member's Support Non-Profit Organization, and is NOT affiliated with any governmental agency or department.