Share Email Print

Proceedings Paper

Comparison of object-oriented with pixel-based classification techniques on urban classification using TM and IKONOS imagery
Author(s): Peifa Wang; Xuezhi Feng; Shuhe Zhao; Pengfeng Xiao; Chunyan Xu
Format Member Price Non-Member Price
PDF $14.40 $18.00
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Paper Abstract

During the last decades, researchers have mainly focused on improving of the pixel-based classification methods and their applications. As the image resolution improved, it can't get good classification result. In order to overcome this problem, the object-oriented approaches are introduced. In this paper, two methods were compared on urban area. A part of Nanjing city in china was selected as study area; TM and IKONOS imagery were employed. Three pixel-based classification methods, the maximum likelihood, ISODATA (Iterative Self-Organizing Data Analysis Technique), minimum distance method, and an object-oriented technique, the nearest neighbor method, were used to classify image, and evaluate the result. For TM imagery, the accuracy assessment of the results showed that the object-oriented classification approach couldn't get better classification result comparing to the pixel-based classification method, the salt-pepper phenomena of the pixel-based classification result images were not obvious. For IKONOS imagery, classification results provided by the object-oriented classification method were better than the pixel-based classification approaches. So, for urban classification using TM imagery, the traditional classification method could be used to get classification information and an acceptable result could be acquired. But when the IKONOS imagery was used to investigate the urban class, the object-oriented method could find the expected result.

Paper Details

Date Published: 26 July 2007
PDF: 8 pages
Proc. SPIE 6752, Geoinformatics 2007: Remotely Sensed Data and Information, 67522J (26 July 2007); doi: 10.1117/12.760759
Show Author Affiliations
Peifa Wang, Nanjing Univ. (China)
Xuezhi Feng, Nanjing Univ. (China)
Shuhe Zhao, Nanjing Univ. (China)
Pengfeng Xiao, Nanjing Univ. (China)
Chunyan Xu, Nanjing Univ. (China)

Published in SPIE Proceedings Vol. 6752:
Geoinformatics 2007: Remotely Sensed Data and Information

© SPIE. Terms of Use
Back to Top