Title :
Multilayer perceptron classification for ENVISAT-ASAR imagery
Author :
Zhu, Feiya ; Guo, Huadong ; Dong, Qing ; Wang, Changlin
Author_Institution :
Inst. of Remote Sensing Applications, Chinese Acad. of Sci., Beijing
Abstract :
This paper describes the application of neural networks to targets classification from multi-polarization ENVISAT-ASAR imagery. The used neural network is multilayer perception (MLP) with fast learning (FL), which is fully interconnected network. Accordingly, the training data sets may be taken from a known truth data in the ground. And finally, the results of proposed method are compared with that of the other classification ones, the in situ test data are from Zhaoqing in Guangdong Province of China
Keywords :
geophysical signal processing; geophysical techniques; image classification; learning (artificial intelligence); multilayer perceptrons; remote sensing by radar; synthetic aperture radar; China; Guangdong Province; Zhaoqing; fast learning; fully interconnected network; in situ test data; known truth data; multilayer perceptron classification; multipolarization ENVISAT-ASAR imagery; neural networks; target classification; Biological neural networks; Mathematical model; Multi-layer neural network; Multilayer perceptrons; Neural networks; Polynomials; Probability density function; Remote sensing; Testing; Training data;
Conference_Titel :
Geoscience and Remote Sensing Symposium, 2004. IGARSS '04. Proceedings. 2004 IEEE International
Conference_Location :
Anchorage, AK
Print_ISBN :
0-7803-8742-2
DOI :
10.1109/IGARSS.2004.1370348
