Comparison and visualization of feature space behaviour of statistical and neural classifiers of satellite imagery
Statistical and neural classifiers of satellite imagery have been compared using sets of training and verification pixels extracted from the Landsat TM imagery of Portugal. Visualisation of class decision boundaries in feature space has been used as a means of gaining insight into the classification processes. Results show that the maximum likelihood classifier and the neural network both generate their decision boundaries as intersections of equiprobability surfaces in feature space.
Bibliographic Reference: Paper presented: International Geoscience and Remote Sensing Symposium : IGARSS '94, Pasadena (US), August 8-12, 1994
Availability: Available from (1) as Paper EN 38335 ORA
Record Number: 199410650 / Last updated on: 1994-11-28
Original language: en
Available languages: en