Supervised Majority Choice Image Classification

Author: O. Conrad (c) 2023
Menu: Imagery | Classification

Description

The majority choice tool for supervised image classification runs the selected classification tools using standard settings and takes for each pixel of the resulting classification the class that has been identified most often by the individual classifiers. Random Forest is not selected by default because it generates randomized results with more or less strong differences each time it is run. K-Nearest Neighbours and Artifical Neural Network have been excluded from default, because using them might be a bit more time consuming.

Parameters

	Name	Type	Identifier	Description	Constraints
Input	Features	grid list, input	`FEATURES`	-	-
	Training Areas	shapes, input	`TRAIN_AREAS`	-	-
	Training Samples	table, input	`TRAIN_SAMPLES`	-	-
Output	Majority Choice	grid, output	`CLASSES`	-	-
	Majority Count	grid, output, optional	`MAJORITY_COUNT`	-	-
	Classes Count	grid, output, optional	`NUNIQUES`	-	-
Options	Normalize	boolean	`NORMALIZE`	-	Default: 0
	Training	choice	`MODEL_TRAIN`	-	Available Choices: [0] training areas [1] training samples Default: 0
	Class Identifier	table field	`TRAIN_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: 30.000000
	Grid System	grid system	`GRID_SYSTEM`	-	-
	Unambiguous	boolean	`UNAMBIGUOUS`	Do not classify a pixel if more than one class reaches the same majority count for it.	Default: 0
	Parallel Epiped	boolean	`CLASSIFY_BOX`	-	Default: 1
	Minimum Distance	boolean	`CLASSIFY_MINDIST`	-	Default: 1
	Mahalonobis Distance	boolean	`CLASSIFY_MAHALONOBIS`	-	Default: 1
	Maximum Likelihood	boolean	`CLASSIFY_MAXLIKE`	-	Default: 1
	Spectral Angle Mapping	boolean	`CLASSIFY_SAM`	-	Default: 1
	Normal Bayes	boolean	`CLASSIFY_BAYES`	-	Default: 1
	Decision Tree	boolean	`CLASSIFY_DT`	-	Default: 1
	Random Forest	boolean	`CLASSIFY_RF`	-	Default: 0
	Support Vector Machine	boolean	`CLASSIFY_SVM`	-	Default: 1
	K-Nearest Neighbours	boolean	`CLASSIFY_KNN`	-	Default: 0
	Artificial Neural Network	boolean	`CLASSIFY_ANN`	-	Default: 0

Command Line

Usage: saga_cmd imagery_classification classify_majority [-FEATURES ] [-NORMALIZE ] [-MODEL_TRAIN ] [-TRAIN_AREAS ] [-TRAIN_CLASS ] [-TRAIN_BUFFER ] [-TRAIN_SAMPLES ] [-CLASSES ] [-MAJORITY_COUNT ] [-NUNIQUES ] [-UNAMBIGUOUS ] [-CLASSIFY_BOX ] [-CLASSIFY_MINDIST ] [-CLASSIFY_MAHALONOBIS ] [-CLASSIFY_MAXLIKE ] [-CLASSIFY_SAM ] [-CLASSIFY_BAYES ] [-CLASSIFY_DT ] [-CLASSIFY_RF ] [-CLASSIFY_SVM ] [-CLASSIFY_KNN ] [-CLASSIFY_ANN ]
  -FEATURES:            	Features
	grid list, input
  -NORMALIZE:           	Normalize
	boolean
	Default: 0
  -MODEL_TRAIN:         	Training
	choice
	Available Choices:
	[0] training areas
	[1] training samples
	Default: 0
  -TRAIN_AREAS:         	Training Areas
	shapes, input
  -TRAIN_CLASS:         	Class Identifier
	table field
  -TRAIN_BUFFER:     	Buffer Size
	floating point number
	Minimum: 0.000000
	Default: 30.000000
  -TRAIN_SAMPLES:       	Training Samples
	table, input
  -CLASSES:             	Majority Choice
	grid, output
  -MAJORITY_COUNT:      	Majority Count
	grid, output, optional
  -NUNIQUES:            	Classes Count
	grid, output, optional
  -UNAMBIGUOUS:         	Unambiguous
	boolean
	Default: 0
  -CLASSIFY_BOX:        	Parallel Epiped
	boolean
	Default: 1
  -CLASSIFY_MINDIST:    	Minimum Distance
	boolean
	Default: 1
  -CLASSIFY_MAHALONOBIS:	Mahalonobis Distance
	boolean
	Default: 1
  -CLASSIFY_MAXLIKE:    	Maximum Likelihood
	boolean
	Default: 1
  -CLASSIFY_SAM:        	Spectral Angle Mapping
	boolean
	Default: 1
  -CLASSIFY_BAYES:      	Normal Bayes
	boolean
	Default: 1
  -CLASSIFY_DT:         	Decision Tree
	boolean
	Default: 1
  -CLASSIFY_RF:         	Random Forest
	boolean
	Default: 0
  -CLASSIFY_SVM:        	Support Vector Machine
	boolean
	Default: 1
  -CLASSIFY_KNN:        	K-Nearest Neighbours
	boolean
	Default: 0
  -CLASSIFY_ANN:        	Artificial Neural Network
	boolean
	Default: 0