GRASS logo

NAME

i.pr.features - Module to process training data for feature extration. i.pr: Pattern Recognition environment for image processing. Includes kNN, Decision Tree and SVM classification techniques. Also includes cross-validation and bagging methods for model validation.

KEYWORDS

imagery, image processing, pattern recognition

SYNOPSIS

i.pr.features
i.pr.features help
i.pr.features [-s] training=string[,string,...] features=string [normalize=integer[,integer,...]] [mean=integer[,integer,...]] [variance=integer[,integer,...]] [prin_comp=integer[,integer,...]] [class_pc=integer[,integer,...]] [standardize=integer[,integer,...]] [--verbose] [--quiet]

Flags:

-s
Print site file containing coordinates of examples and class labels (from training data) into features option and exit.
--verbose
Verbose module output
--quiet
Quiet module output

Parameters:

training=string[,string,...]
Input files (max 50) containing training data.
2 formats are currently supported:
1) GRASS_data (output of i.pr_training)
2) TABLE_data.
features=string
Name of the output file containing the features.
normalize=integer[,integer,...]
Numbers of the layers to be normalized.
mean=integer[,integer,...]
Numbers of the layers on which to compute the mean value.
variance=integer[,integer,...]
Numbers of the layers on which to compute the variance.
prin_comp=integer[,integer,...]
Numbers of the layers on which to compute the principal components.
class_pc=integer[,integer,...]
Classes of the data to be used for computing the principal components.
If not set, all the examples will be used.
standardize=integer[,integer,...]
Numbers of features to be standardize.
WARNING: not related to the number of layers.

DESCRIPTION

i.pr.features This module is used to preprocess and extract the training features. It is necessary to specify the training file, which contains the names of the raster maps (explanatory variables) to be used in subsequent modules. The training file can either be the output from from i.pr.training (Recommended) or an ascii file containing the names of rasters.

This module allows for the calculation of a range of statistics pertaining to the explanatory variables, which include the mean and variance. In addition the features can be normalized to a similar scale. In each case, it is possible to specify the numbers of features for which these statistics should be computed (i.e. number in list). There is also the possibility to compute principal components for the explanatory variables. The default calculates them for all layers, or else only on selected classes specified by 'class_pc'. Variables can be standardised using parameter 'standardize', this is linked to the features previously calculated and not layers in the training file.

NOTES

SEE ALSO

i.pr.training
i.pca

AUTHORS

Stefano Merler, FBK, Trento, Italy
Documentation: Daniel McInerney (daniel.mcinerney ucd.ie)

Last changed: $Date$


Main index - imagery index - Full index

© 2003-2016 GRASS Development Team