Title :
A method to classify multispectral images
Author_Institution :
Inst. of Photogrammetry & Remote Sensing, Karlsruhe Univ., Germany
Abstract :
The author´s method performs an unsupervised clustering of the feature vectors. In order to classify the clusters they are compared with already classified clusters in a database using several metrics. The metrics take into consideration the positions and the shapes of the clusters. The author applies this method to Landsat-TM images to make land-use classifications and significantly better results are obtained than the maximum likelihood classification method
Keywords :
geophysical signal processing; geophysical techniques; image classification; optical information processing; remote sensing; Landsat-TM image; feature vector; geophysical measurement technique; land surface; land-use; metrics; multidimensional processing; multispectral image classification; multispectral remote sensing; optical imaging; terrain mapping; unsupervised clustering; visible IR infrared; Clustering algorithms; Hypercubes; Image databases; Maximum likelihood estimation; Multispectral imaging; Pixel; Remote sensing; Satellites; Shape; Spatial databases; Tree graphs;
Conference_Titel :
Geoscience and Remote Sensing Symposium, 1995. IGARSS '95. 'Quantitative Remote Sensing for Science and Applications', International
Conference_Location :
Firenze
Print_ISBN :
0-7803-2567-2
DOI :
10.1109/IGARSS.1995.521172
