Title :
Fusion of contextual classification and the existing classification result
Author_Institution :
Hiroshima Univ., Higashi-Hiroshima, Japan
Abstract :
We consider discriminant analysis of geo-spatial data under the situation such that a classified result is already available. We propose a fusion method of a contextual classification result based on Markov random fields and the existing result. Then, our procedure is applied to the benchmark data set for discrimination provided by IEEE GRSS Data Fusion Committee.
Keywords :
geophysical signal processing; geophysical techniques; image classification; remote sensing; sensor fusion; terrain mapping; Markov random fields; contextual classification; data fusion; discriminant analysis; fusion method; geo-spatial data; geophysical measurement technique; image classification; image fusion; image processing; image segmentation; land surface; remote sensing; terrain mapping; Benchmark testing; Data analysis; Gaussian distribution; Image segmentation; Iterative algorithms; Markov random fields; Parameter estimation; Probability;
Conference_Titel :
Geoscience and Remote Sensing Symposium, 2002. IGARSS '02. 2002 IEEE International
Print_ISBN :
0-7803-7536-X
DOI :
10.1109/IGARSS.2002.1024947
