ISODATA Clustering for Grids

Author: O.Conrad (c) 2016
Menu: Imagery | Classification | Unsupervised

Description

This tool executes the Isodata unsupervised classification - clustering algorithm. Isodata stands for Iterative Self-Organizing Data Analysis Techniques. This is a more sophisticated algorithm which allows the number of clusters to be automatically adjusted during the iteration by merging similar clusters and splitting clusters with large standard deviations. The tool is based on Christos Iosifidis' Isodata implementation.

References

isodata.c (Christos Iosifidis)
A Fast Implementation of the ISODATA Clustering Algorithm
Memarsadeghi, N., Mount, D. M., Netanyahu, N. S., Le Moigne, J. (2007): A Fast Implementation of the ISODATA Clustering Algorithm. International Journal of Computational Geometry and Applications, 17, 71-103. online.

Parameters

	Name	Type	Identifier	Description	Constraints
Input	Features	grid list, input	`FEATURES`	-	-
Output	Clusters	grid, output	`CLUSTER`	-	-
Output	Statistics	table, output	`STATISTICS`	-	-
Options	Grid System	grid system	`PARAMETERS_GRID_SYSTEM`	-	-
	Normalize	boolean	`NORMALIZE`	-	Default: 0
	Maximum Number of Iterations	integer number	`ITERATIONS`	-	Minimum: 3 Default: 20
	Initial Number of Clusters	integer number	`CLUSTER_INI`	-	Minimum: 0 Default: 5
	Maximum Number of Clusters	integer number	`CLUSTER_MAX`	-	Minimum: 3 Default: 16
	Minimum Number of Samples in a Cluster	integer number	`SAMPLES_MIN`	-	Minimum: 2 Default: 5
	Update Colors from Features	boolean, GUI	`RGB_COLORS`	Use the first three features in list to obtain blue, green, red components for class colour in look-up table.	Default: 0
	Start Partition	choice	`INITIALIZE`	-	Available Choices: [0] random [1] periodical [2] keep values Default: 0

Command Line


Usage: saga_cmd imagery_isocluster 0 [-FEATURES ] [-CLUSTER ] [-STATISTICS ] [-NORMALIZE ] [-ITERATIONS ] [-CLUSTER_INI ] [-CLUSTER_MAX ] [-SAMPLES_MIN ] [-INITIALIZE ]
  -FEATURES:   	Features
	grid list, input
  -CLUSTER:    	Clusters
	grid, output
  -STATISTICS: 	Statistics
	table, output
  -NORMALIZE:  	Normalize
	boolean
	Default: 0
  -ITERATIONS: 	Maximum Number of Iterations
	integer number
	Minimum: 3
	Default: 20
  -CLUSTER_INI:	Initial Number of Clusters
	integer number
	Minimum: 0
	Default: 5
  -CLUSTER_MAX:	Maximum Number of Clusters
	integer number
	Minimum: 3
	Default: 16
  -SAMPLES_MIN:	Minimum Number of Samples in a Cluster
	integer number
	Minimum: 2
	Default: 5
  -INITIALIZE: 	Start Partition
	choice
	Available Choices:
	[0] random
	[1] periodical
	[2] keep values
	Default: 0