Research on remote sensing classification based on improved Kohonen neural network

Author

Luo, Xiaobo ; Liu, Qinhuo ; Liu, Qiang ; Xia, Ying

Author_Institution

Inst. of Remote Sensing Applic., Chinese Acad. of Sci., Beijing, China

Volume

4

fYear

2010

fDate

16-18 April 2010

Abstract

In this paper, we used max-min distance method to determine Kohonen Network´ initial weights, as well as introduced energy function as the convergence condition of network, and as such improved Kohonen network´ unsupervised-learning algorithm. We then used the improved Kohonen´ learning algorithm combined with penalty formula to carry out supervised classification of remote sensing data. Experiments showed that higher accuracy resulted with this improved algorithm.

Keywords

pattern classification; remote sensing; self-organising feature maps; unsupervised learning; Kohonen neural network; energy function; max-min distance method; penalty formula; remote sensing classification; supervised data classification; unsupervised learning algorithm; Application software; Artificial neural networks; Computer science; Convergence; Detectors; Educational institutions; Neural networks; Neurons; Remote sensing; Unsupervised learning; Kohonen neural network; max-min distance means; supervised classification;

fLanguage

English

Publisher

ieee

Conference_Titel

Computer Engineering and Technology (ICCET), 2010 2nd International Conference on

Conference_Location

Chengdu

Print_ISBN

978-1-4244-6347-3

Type

conf

DOI

10.1109/ICCET.2010.5485413

Filename

5485413