CLASSIFICATION METHODS FOR REMOTELY SENSED DATA

Brandt TSO and Paul M. MATHER

Published by Taylor and Francis Ltd (London) and Taylor and Francis Inc (New York)

ISBN 0-415-25909-6 (paperback) (£29)
ISBN 0-415-25908-8 (hardback)   (£65)

Publication date: 19 October, 2001

To order, visit the Taylor and Francis GIS and Remote Sensing Arena web site


Remote sensing is now an integral part of geography, GIS, and cartography, and remotely sensed images are used in research and applications in a range of environmental and planning fields.

One of the most important application of remotely sensed images of the Earth's land surface is the classification or labelling of pixels or groups of pixels - in other words, identifying  image pixels in terms of land cover categories or types that they represent. This is the process of classification.

This book provides a comprehensive survey of image classification techniques. Beginning with a general, introductory review of basic principles, the book aims to  bring together information from a range of sources and set it in the context of basic principles. There is an emphasis on new methods, including the use of artificial neural networks, procedures based on fuzzy theory, techniques of texture quantisation, the use of Markov random fields in modelling context, and the use of multiple classifiers.

Readership: The early chapters provide an extensive introduction for advanced undergraduates, while later chapters address specific topics at a level suitable for post-graduate researchers.

This page provides links to software written in C by the first author of the book, Dr Brandt Tso.

Contact:  Prof. Paul M. Mather, School of Geography, The University of Nottingham, NG7 2RD (England)

Contents:

1.     Remote Sensing in the Optical and Microwave Regions
2.     Pattern Recognition Principles
3.     Pattern Recognition Using Artificial Neural Networks
4.     Methods Based on Fuzzy Set Theory
5.     Texture Quantisation
6.     Modelling Context Using Markov Random Fields
7.     Multi-source Classification

Software


The following downloadable ASCII files contain C implementations of procedures described in the book. They are provided as examples and not as complete or working programs. No warranty or guarantee is given that they will compile or execute on any particular system or with any particular compiler, nor does the author claim that these examples are efficient or effective. In other words - please use them, but we cannot be held responsible for any malfunctions that you or your computer may experience!

©  (2001) Brandt Tso. Any commercial use of this software requires the agreement of Brandt Tso.

Click here to email Brandt Tso.

Note that this software is provided by Brandt Tso and not by Taylor and Francis Ltd.
 

Besag's Iterative Conditional Modes (ICM) procedure
Boundary generate - floating point to generic binary with boundary information
Combine Band Interleave - reads generic binary (BSQ) file, outputs BIL file
Confusion Matrix (plus Kappa)
Correlation coefficient for two images
Divergence measure
Equal probabilities (normalise pixel values)
Evidence combination for use with evidential reasoning
Fuzzy c means clustering
Fuzzy rules - example program
Fuzzy rules - example program using hierarchical rule base
Fuzzy Rules: generate test data file
Generate texture using MAP
Generate test and training data for SNNS (ANN software)
GLCM texture measures
Histogram of image file
Isodata clustering
Lacunarity measure
Mahalanobis distance
Maximum likelihood classification
Mean, covariance and inverse covariance matrix for image set
Test of spectral unmixing
MAP-based classification (Marroquin 1987)
MAR (texture parameters)
MRF - estimate parameters
Pre-processing training data using robust mean and covariance method
Restore noise-fading texture image using MRF
Simulated Annealing
SOM/LVQ classification
Semivariogram and fractal dimension for transect
Subset a generic binary image