Supervised Image Classification

Author: O.Conrad (c) 2005
Menu: Imagery | Classification

Description

Standard methods for supervised image classification, including minimum distance, maximum likelihood, spectral angle mapping. Classifiers can be trained by areas defined through shapes, samples supplied as table records, or statistics previously stored to file.

Parameters

	Name	Type	Identifier	Description	Constraints
Input	Features	grid list, input	`GRIDS`	-	-
	Training Areas	shapes, input	`TRAINING`	-	-
	Training Samples	table, input	`TRAIN_SAMPLES`	Provide a class identifier in the first field followed by sample data corresponding to the input feature grids.	-
Output	Classification	grid, output	`CLASSES`	-	-
	Quality	grid, output, optional	`QUALITY`	Dependent on chosen method, these are distances or probabilities.	-
	Look-up Table	table, output, optional	`CLASSES_LUT`	A reference list of the grid values that have been assigned to the training classes.	-
Options	Normalize	boolean	`NORMALISE`	-	Default: 0
	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
	Grid System	grid system	`GRID_SYSTEM`	-	-
	Training	choice	`TRAIN_WITH`	-	Available Choices: [0] training areas [1] training samples [2] load from file Default: 0
	Class Identifier	table field	`TRAINING_CLASS`	-	-
	Buffer Size	floating point number	`TRAIN_BUFFER`	For non-polygon type training areas, creates a buffer with a diameter of specified size.	Minimum: 0.000000 Default: 1.000000
	Load Statistics from File...	file path	`FILE_LOAD`	-	-
	Save Statistics to File...	file path	`FILE_SAVE`	-	-
	Method	choice	`METHOD`	-	Available Choices: [0] Binary Encoding [1] Parallelepiped [2] Minimum Distance [3] Mahalanobis Distance [4] Maximum Likelihood [5] Spectral Angle Mapping Default: 2
	Distance Threshold	floating point number	`THRESHOLD_DIST`	Let pixel stay unclassified, if minimum euclidian or mahalanobis distance is greater than threshold.	Minimum: 0.000000 Default: 0.000000
	Spectral Angle Threshold (Degree)	floating point number	`THRESHOLD_ANGLE`	Let pixel stay unclassified, if spectral angle distance is greater than threshold.	Minimum: 0.000000 Maximum: 90.000000 Default: 0.000000
	Probability Threshold	floating point number	`THRESHOLD_PROB`	Let pixel stay unclassified, if maximum likelihood probability value is less than threshold.	Minimum: 0.000000 Maximum: 100.000000 Default: 0.000000
	Probability Reference	choice	`RELATIVE_PROB`	-	Available Choices: [0] absolute [1] relative Default: 1

Command Line


Usage: saga_cmd imagery_classification 0 [-GRIDS ] [-NORMALISE ] [-CLASSES ] [-QUALITY ] [-CLASSES_LUT ] [-TRAIN_WITH ] [-TRAINING ] [-TRAINING_CLASS ] [-TRAIN_BUFFER ] [-TRAIN_SAMPLES ] [-FILE_LOAD ] [-FILE_SAVE ] [-METHOD ] [-THRESHOLD_DIST ] [-THRESHOLD_ANGLE ] [-THRESHOLD_PROB ] [-RELATIVE_PROB ]
  -GRIDS:             	Features
	grid list, input
  -NORMALISE:         	Normalize
	boolean
	Default: 0
  -CLASSES:           	Classification
	grid, output
  -QUALITY:           	Quality
	grid, output, optional
  -CLASSES_LUT:       	Look-up Table
	table, output, optional
  -TRAIN_WITH:        	Training
	choice
	Available Choices:
	[0] training areas
	[1] training samples
	[2] load from file
	Default: 0
  -TRAINING:          	Training Areas
	shapes, input
  -TRAINING_CLASS:    	Class Identifier
	table field
  -TRAIN_BUFFER:   	Buffer Size
	floating point number
	Minimum: 0.000000
	Default: 1.000000
  -TRAIN_SAMPLES:     	Training Samples
	table, input
  -FILE_LOAD:         	Load Statistics from File...
	file path
  -FILE_SAVE:         	Save Statistics to File...
	file path
  -METHOD:            	Method
	choice
	Available Choices:
	[0] Binary Encoding
	[1] Parallelepiped
	[2] Minimum Distance
	[3] Mahalanobis Distance
	[4] Maximum Likelihood
	[5] Spectral Angle Mapping
	Default: 2
  -THRESHOLD_DIST: 	Distance Threshold
	floating point number
	Minimum: 0.000000
	Default: 0.000000
  -THRESHOLD_ANGLE:	Spectral Angle Threshold (Degree)
	floating point number
	Minimum: 0.000000
	Maximum: 90.000000
	Default: 0.000000
  -THRESHOLD_PROB: 	Probability Threshold
	floating point number
	Minimum: 0.000000
	Maximum: 100.000000
	Default: 0.000000
  -RELATIVE_PROB:     	Probability Reference
	choice
	Available Choices:
	[0] absolute
	[1] relative
	Default: 1