User Manual

The GUI

Toolbar

Data Sources

Data Views




Applications / Tools

Image Statistics

Reclassify

Scatterplot

Virtual Raster Builder

imageMath Calculator




Processing Data Types

General processing schema

Todo

add figure

Raster

Supported output filetypes:

  • .tif

Mask

Mask files can be used to exclude pixels from certain processes. This allows to constrain operations on regions of interest only and to reduce computational costs. Any GDAL/OGR readable raster or vector file can be interpreted as a boolean mask.

  • In case of a vector, all pixels covered by features are interpreted as True, all other pixels as False.
  • In case of a raster, all pixels that are equal to the no data value (default is 0) are interpreted as False, all other pixels as True. Multiband rasters are first evaluated band wise. The final mask for a given pixel is True, if all band wise masks for that pixel are True. It is assumed that all pixels with a value of zero mark a position that is to be masked and neglected during a specific operation.

Classification

Regression

Fraction

Spectral Library

Sample

Fit




Processing Algorithms

Accuracy Assessment

Classification Performance

Assesses the performance of a classification.

Prediction

Specify classification raster be evaluated

Reference

Specify reference classification raster (i.e. ground truth).

HTML Report

Specify output path for HTML report file (.html).


Clustering Performance

Assesses the performance of a clusterer.

Prediction

Specify clustering raster to be evaluated.

Reference

Specify reference clustering raster (i.e. ground truth).

HTML Report

Specify output path for HTML report file (.html).


ClassProbability Performance

Assesses the performance of class probabilities in terms of AUC and ROC curves.

Prediction

Specify class probability raster to be evaluated.

Reference

Specify reference classification raster (i.e. ground truth).

HTML Report

Specify output path for HTML report file (.html).


Regression Performance

Assesses the performance of a regression.

Prediction

Specify regression raster to be evaluated.

Reference

Specify reference regression raster (i.e. ground truth).

HTML Report

Specify output path for HTML report file (.html).


Auxilliary

ClassDefinition from Raster

Creates a Class Definition string from a classification input raster for the usage in other EnMAP-Box algorithms (e.g. ‘Classification from Vector’). See Log window for result.

Raster

Specify raster with defined class definition, e.g. classification or class probability raster


Create additional Testdata

Based on the testdata additional datasets will be created using existing EnMAP-Box algorithms with predefined settings.

LandCover L2 Classification

Specify output path for LandCover L2 Classification.

LandCover L2 ClassProbability

Specify output path for LandCover L2 ClassProbability.

Output Sample

Specify output path for sample (.pkl).

Output ClassificationSample

Specify output path for sample (.pkl).

Output ClassProbabilitySample

Specify output path for sample (.pkl).

Output RegressionSample

Specify output path for sample (.pkl).


Open Testdata

Opens testdata into current QGIS project (LandCov_BerlinUrbanGradient.shp, HighResolution_BerlinUrbanGradient.bsq, EnMAP_BerlinUrbanGradient.bsq, SpecLib_BerlinUrbanGradient.sli).

EnMAP (30m; 177 bands)

File name: EnMAP_BerlinUrbanGradient.bsq

Simulated EnMAP data (based on 3.6m HyMap imagery) acquired in August 2009 over south eastern part of Berlin covering an area of 4.32 km^2 (2.4 x 1.8 km). It has a spectral resolution of 177 bands and a spatial resolution of 30m.

HyMap (3.6m; Blue, Green, Red, NIR bands)

File name: HighResolution_BerlinUrbanGradient.bsq

HyMap image acquired in August 2009 over south eastern part of Berlin covering an area of 4.32 km² (2.4 x 1.8 km). This dataset was reduced to 4 bands (0.483, 0.558, 0.646 and 0.804 micrometers). The spatial resolution is 3.6m.

LandCover Layer

File name: LandCov_BerlinUrbanGradient.shp

Polygon shapefile containing land cover information on two classification levels. Derived from very high resolution aerial imagery and cadastral datasets.

Level 1 classes: Impervious; Other; Vegetation; Soil

Level 2 classes: Roof; Low vegetation; Other; Pavement; Tree; Soil

ENVI Spectral Library

File name: SpecLib_BerlinUrbanGradient.sli

Spectral library with 75 spectra (material level, level 2 and level 3 class information)


Scale Sample Features

Scales the features of a sample by a user defined factor (can be used for matching datasets). Use case: A sample from a spectral library should be used for classifying a raster. The spectral library sample has float surface reflectance values between 0 and 1 and the raster integer surface reflectances between 0 and 1000. In order to match the datasets, you can rescale the sample by a factor of 1000.

Sample

Specify path to sample file (.pkl).

Scale factor

Scale factor that is applied to all features.

Output Sample

Specify output path for sample (.pkl).


Unique Values from Vector Attribute

This algorithm returns unique values from vector attributes as a list, which is also usable as Class Definition in other algorithms. The output will be shown in the log window and can the copied from there accordingly.

Vector

Specify input vector.

Field

Specify field of vector layer for which unique values should be derived.


Classification

Fit GaussianProcessClassifier

Fits Gaussian Process Classifier. See Gaussian Processes for further information.

ClassificationSample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. See GaussianProcessClassifier for information on different parameters.

Output Classifier

Specifiy output path for the classifier (.pkl). This file can be used for applying the classifier to an image using ‘Classification -> Predict Classification’ and ‘Classification -> Predict ClassProbability’.


Fit LinearSVC

Fits a linear Support Vector Classification. Input data will be scaled and grid search is used for model selection.

ClassificationSample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. For information on different parameters have a look at LinearSVC. See GridSearchCV for information on grid search and StandardScaler for scaling.

Output Classifier

Specifiy output path for the classifier (.pkl). This file can be used for applying the classifier to an image using ‘Classification -> Predict Classification’ and ‘Classification -> Predict ClassProbability’.


Fit RandomForestClassifier

Fits a Random Forest Classifier

ClassificationSample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. See RandomForestClassifier for information on different parameters. If this code is not altered, scikit-learn default settings will be used. ‘Hint: you might want to alter e.g. the n_estimators value (number of trees), as the default is 10. So the line of code might be altered to ‘estimator = RandomForestClassifier(n_estimators=100).’

Output Classifier

Specifiy output path for the classifier (.pkl). This file can be used for applying the classifier to an image using ‘Classification -> Predict Classification’ and ‘Classification -> Predict ClassProbability’.


Fit SVC

Fits a Support Vector Classification. Input data will be scaled and grid search is used for model selection.

ClassificationSample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. For information on different parameters have a look at SVC. See GridSearchCV for information on grid search and StandardScaler for scaling.

Output Classifier

Specifiy output path for the classifier (.pkl). This file can be used for applying the classifier to an image using ‘Classification -> Predict Classification’ and ‘Classification -> Predict ClassProbability’.


Predict Classification

Applies a classifier to a raster.

Raster

Select raster file which should be classified.

Mask

Specified vector or raster is interpreted as a boolean mask.

In case of a vector, all pixels covered by features are interpreted as True, all other pixels as False.

In case of a raster, all pixels that are equal to the no data value (default is 0) are interpreted as False, all other pixels as True.Multiband rasters are first evaluated band wise. The final mask for a given pixel is True, if all band wise masks for that pixel are True.

Classifier

Select path to a classifier file (.pkl).

Output Classification

Specify output path for classification raster.


Predict ClassProbability

Applies a classifier to a raster.

Raster

Specify input raster.

Mask

Specified vector or raster is interpreted as a boolean mask.

In case of a vector, all pixels covered by features are interpreted as True, all other pixels as False.

In case of a raster, all pixels that are equal to the no data value (default is 0) are interpreted as False, all other pixels as True.Multiband rasters are first evaluated band wise. The final mask for a given pixel is True, if all band wise masks for that pixel are True.

Classifier

Select path to a classifier file (.pkl).

Prediction

Specify output path for raster.


Clustering

Fit AffinityPropagation

Fits a Affinity Propagation clusterer (input data will be scaled).

Sample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. For information on different parameters have a look at AffinityPropagation. See StandardScaler for information on scaling

Output Clusterer

Specifiy output path for the clusterer (.pkl). This file can be used for applying the clusterer to an image using ‘Clustering -> Predict Clustering’.


Fit Birch

Fits a Birch clusterer (input data will be scaled).

Sample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. For information on different parameters have a look at Birch. See StandardScaler for information on scaling

Output Clusterer

Specifiy output path for the clusterer (.pkl). This file can be used for applying the clusterer to an image using ‘Clustering -> Predict Clustering’.


Fit KMeans

Fits a KMeans clusterer (input data will be scaled).

Sample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. For information on different parameters have a look at KMeans. See StandardScaler for information on scaling

Output Clusterer

Specifiy output path for the clusterer (.pkl). This file can be used for applying the clusterer to an image using ‘Clustering -> Predict Clustering’.


Fit MeanShift

Fits a MeanShift clusterer (input data will be scaled).

Sample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. For information on different parameters have a look at MeanShift. See StandardScaler for information on scaling

Output Clusterer

Specifiy output path for the clusterer (.pkl). This file can be used for applying the clusterer to an image using ‘Clustering -> Predict Clustering’.


Predict Clustering

Applies a clusterer to a raster.

Raster

Select raster file which should be clustered.

Mask

Specified vector or raster is interpreted as a boolean mask.

In case of a vector, all pixels covered by features are interpreted as True, all other pixels as False.

In case of a raster, all pixels that are equal to the no data value (default is 0) are interpreted as False, all other pixels as True.Multiband rasters are first evaluated band wise. The final mask for a given pixel is True, if all band wise masks for that pixel are True.

Clusterer

Select path to a clusterer file (.pkl).

Clustering

Specify output path for classification raster.


Create Raster

Classification from ClassProbability

Creates classification from class probability. Winner class is equal to the class with maximum class probability.

ClassProbability

Specify input raster.

Minimal overall coverage

Mask out all pixels that have an overall coverage less than the specified value. This controls how edges between labeled and no data regions are treated.

Minimal winner class coverage

Mask out all pixels that have a coverage of the predominant class less than the specified value. This controls pixel purity.

Output Classification

Specify output path for classification raster.


Classification from Vector

Creates a classification from a vector field with class ids.

PixelGrid

Specify input raster.

Vector

Specify input vector.

Class id attribute

Vector field specifying the class ids.

Class Definition

Enter a class definition, e.g.:

ClassDefinition(names=[‘Urban’, ‘Forest’, ‘Water’], colors=[‘red’, ‘#00FF00’, (0, 0, 255)])

For supported named colors see the W3C recognized color keyword names.

Minimal overall coverage

Mask out all pixels that have an overall coverage less than the specified value. This controls how edges between labeled and no data regions are treated.

Minimal winner class coverage

Mask out all pixels that have a coverage of the predominant class less than the specified value. This controls pixel purity.

Oversampling factor

Defines the degree of detail by which the class information given by the vector is rasterized. An oversampling factor of 1 (default) simply rasterizes the vector on the target pixel grid.An oversampling factor of 2 will rasterize the vector on a target pixel grid with resolution twice as fine.An oversampling factor of 3 will rasterize the vector on a target pixel grid with resolution three times as fine, … and so on.

Mind that larger values are always better (more accurate), but depending on the inputs, this process can be quite computationally intensive, when a higher factor than 1 is used.

Output Classification

Specify output path for classification raster.


ClassProbability from Classification

Derive (binarized) class probabilities from a classification.

Classification

Specify input raster.

Output ClassProbability

Specify output path for class probability raster.


ClassProbability from Vector

Derives class probability raster from a vector file with sufficient class information.

PixelGrid

Specify input raster.

Vector

Specify input vector.

Class id attribute

Vector field specifying the class ids.

Class Definition

Enter a class definition, e.g.:

ClassDefinition(names=[‘Urban’, ‘Forest’, ‘Water’], colors=[‘red’, ‘#00FF00’, (0, 0, 255)])

For supported named colors see the W3C recognized color keyword names.

Minimal overall coverage

Mask out all pixels that have an overall coverage less than the specified value. This controls how edges between labeled and no data regions are treated.

Minimal winner class coverage

Mask out all pixels that have a coverage of the predominant class less than the specified value. This controls pixel purity.

Oversampling factor

Defines the degree of detail by which the class information given by the vector is rasterized. An oversampling factor of 1 (default) simply rasterizes the vector on the target pixel grid.An oversampling factor of 2 will rasterize the vector on a target pixel grid with resolution twice as fine.An oversampling factor of 3 will rasterize the vector on a target pixel grid with resolution three times as fine, … and so on.

Mind that larger values are always better (more accurate), but depending on the inputs, this process can be quite computationally intensive, when a higher factor than 1 is used.

Output ClassProbability

Specify output path for class probability raster.


Raster from Vector

Converts vector to raster (using gdal rasterize).

PixelGrid

Specify input raster.

Vector

Specify input vector.

Init Value

Pre-initialization value for the output raster before burning. Note that this value is not marked as the nodata value in the output raster.

Burn Value

Fixed value to burn into each pixel, which is covered by a feature (point, line or polygon).

Burn Attribute

Specify numeric vector field to use as burn values.

All touched

Enables the ALL_TOUCHED rasterization option so that all pixels touched by lines or polygons will be updated, not just those on the line render path, or whose center point is within the polygon.

Filter SQL

Create SQL based feature selection, so that only selected features will be used for burning.

Example: Level_2 = ‘Roof’ will only burn geometries where the Level_2 attribute value is equal to ‘Roof’, others will be ignored. This allows you to subset the vector dataset on-the-fly.

Data Type

Specify output datatype.

No Data Value

Specify output no data value.

Output Raster

Specify output path for raster.


Create Sample

ClassificationSample from ENVI Spectral Library

Derive ClassificationSample from ENVI Spectral Library.

ENVI Spectral Library

Select path to an ENVI (e.g. .sli or .esl).

ClassDefinition prefix

Class definition prefixes allow the selection of a specific class definition (i.e. ‘class names’ and ‘class lookup’) and class mapping (i.e. ‘class spectra names’) stored in the spectral library .hdr file).

For example, inside the EnMAP-Box testdata spectral library, the prefixes ‘level 1’ and ‘level 2’ are defined.

Output ClassificationSample

Specify output path for sample (.pkl).


ClassificationSample from ClassProbabilitySample

Derive ClassificationSample from ClassProbabilitySample. Winner class is selected by the maximum probability decision.

ClassProbabilitySample

Specify path to sample file (.pkl).

Minimal overall coverage

Mask out all pixels that have an overall coverage less than the specified value. This controls how edges between labeled and no data regions are treated.

Minimal winner class coverage

Mask out all pixels that have a coverage of the predominant class less than the specified value. This controls pixel purity.

Output ClassificationSample

Specify output path for sample (.pkl).


ClassificationSample from Raster and ClassProbability

Derives classification sample from raster and class probability raster.

Raster

Specify input raster.

ClassProbability

Specify input raster.

Mask

Specified vector or raster is interpreted as a boolean mask.

In case of a vector, all pixels covered by features are interpreted as True, all other pixels as False.

In case of a raster, all pixels that are equal to the no data value (default is 0) are interpreted as False, all other pixels as True.Multiband rasters are first evaluated band wise. The final mask for a given pixel is True, if all band wise masks for that pixel are True.

Minimal overall coverage

Mask out all pixels that have an overall coverage less than the specified value. This controls how edges between labeled and no data regions are treated.

Minimal winner class coverage

Mask out all pixels that have a coverage of the predominant class less than the specified value. This controls pixel purity.

Output ClassificationSample

Specify output path for sample (.pkl).


ClassificationSample from Raster and Vector

Derives classification sample from raster and vector.

Raster

Specify input raster.

Vector

Specify input vector.

Class id attribute

Vector field specifying the class ids.

Class Definition

Enter a class definition, e.g.:

ClassDefinition(names=[‘Urban’, ‘Forest’, ‘Water’], colors=[‘red’, ‘#00FF00’, (0, 0, 255)])

For supported named colors see the W3C recognized color keyword names.

Minimal overall coverage

Mask out all pixels that have an overall coverage less than the specified value. This controls how edges between labeled and no data regions are treated.

Minimal winner class coverage

Mask out all pixels that have a coverage of the predominant class less than the specified value. This controls pixel purity.

Oversampling factor

Defines the degree of detail by which the class information given by the vector is rasterized. An oversampling factor of 1 (default) simply rasterizes the vector on the target pixel grid.An oversampling factor of 2 will rasterize the vector on a target pixel grid with resolution twice as fine.An oversampling factor of 3 will rasterize the vector on a target pixel grid with resolution three times as fine, … and so on.

Mind that larger values are always better (more accurate), but depending on the inputs, this process can be quite computationally intensive, when a higher factor than 1 is used.

Mask

Specified vector or raster is interpreted as a boolean mask.

In case of a vector, all pixels covered by features are interpreted as True, all other pixels as False.

In case of a raster, all pixels that are equal to the no data value (default is 0) are interpreted as False, all other pixels as True.Multiband rasters are first evaluated band wise. The final mask for a given pixel is True, if all band wise masks for that pixel are True.

Output ClassificationSample

Specify output path for sample (.pkl).


ClassProbabilitySample from synthetically mixed ClassificationSample

Derives a class probability sample by synthetically mixing (pure) spectra from a ClassificationSample.

ClassificationSample

Specify path to sample file (.pkl).

n

Total number of samples to be generated.

Likelihood for mixing complexity 2

Specifies the probability of mixing spectra from 2 classes.

Likelihood for mixing complexity 3

Specifies the probability of mixing spectra from 3 classes.

Class likelihoods

Specifies the likelihoods for drawing spectra from individual classes.

In case of ‘equalized’, all classes have the same likelihhod to be drawn from.

In case of ‘proportional’, class likelihoods scale with their sizes.

Output ClassProbabilitySample

Specify output path for sample (.pkl).


ClassificationSample from Raster and Classification

Derives a classification sample from raster (defines the grid) and classification.

Raster

Specify input raster.

Classification

Specify input raster.

Mask

Specified vector or raster is interpreted as a boolean mask.

In case of a vector, all pixels covered by features are interpreted as True, all other pixels as False.

In case of a raster, all pixels that are equal to the no data value (default is 0) are interpreted as False, all other pixels as True.Multiband rasters are first evaluated band wise. The final mask for a given pixel is True, if all band wise masks for that pixel are True.

Minimal overall coverage

Mask out all pixels that have an overall coverage less than the specified value. This controls how edges between labeled and no data regions are treated.

Minimal winner class coverage

Mask out all pixels that have a coverage of the predominant class less than the specified value. This controls pixel purity.

Output ClassificationSample

Specify output path for sample (.pkl).


ClassProbabilitySample from ClassificationSample

Derives a class probability sample from a classification sample.

ClassificationSample

Specify path to sample file (.pkl).

Output ClassProbabilitySample

Specify output path for sample (.pkl).


ClassProbabilitySample from Raster and Classification

Derives a class probability sample from raster and classification.

Raster

Specify input raster.

Classification

Specify input raster.

Mask

Specified vector or raster is interpreted as a boolean mask.

In case of a vector, all pixels covered by features are interpreted as True, all other pixels as False.

In case of a raster, all pixels that are equal to the no data value (default is 0) are interpreted as False, all other pixels as True.Multiband rasters are first evaluated band wise. The final mask for a given pixel is True, if all band wise masks for that pixel are True.

Output ClassProbabilitySample

Specify output path for sample (.pkl).


ClassProbabilitySample from Raster and ClassProbability

Derives class probability sample from raster and class probability.

Raster

Specify input raster.

ClassProbability

Specify input raster.

Mask

Specified vector or raster is interpreted as a boolean mask.

In case of a vector, all pixels covered by features are interpreted as True, all other pixels as False.

In case of a raster, all pixels that are equal to the no data value (default is 0) are interpreted as False, all other pixels as True.Multiband rasters are first evaluated band wise. The final mask for a given pixel is True, if all band wise masks for that pixel are True.

Output ClassProbabilitySample

Specify output path for sample (.pkl).


ClassProbabilitySample from Raster and Vector

Derives class probability sample from raster and vector.

Raster

Specify input raster.

Vector

Specify input vector.

Class id attribute

Vector field specifying the class ids.

Class Definition

Enter a class definition, e.g.:

ClassDefinition(names=[‘Urban’, ‘Forest’, ‘Water’], colors=[‘red’, ‘#00FF00’, (0, 0, 255)])

For supported named colors see the W3C recognized color keyword names.

Minimal overall coverage

Mask out all pixels that have an overall coverage less than the specified value. This controls how edges between labeled and no data regions are treated.

Minimal winner class coverage

Mask out all pixels that have a coverage of the predominant class less than the specified value. This controls pixel purity.

Oversampling factor

Defines the degree of detail by which the class information given by the vector is rasterized. An oversampling factor of 1 (default) simply rasterizes the vector on the target pixel grid.An oversampling factor of 2 will rasterize the vector on a target pixel grid with resolution twice as fine.An oversampling factor of 3 will rasterize the vector on a target pixel grid with resolution three times as fine, … and so on.

Mind that larger values are always better (more accurate), but depending on the inputs, this process can be quite computationally intensive, when a higher factor than 1 is used.

Mask

Specified vector or raster is interpreted as a boolean mask.

In case of a vector, all pixels covered by features are interpreted as True, all other pixels as False.

In case of a raster, all pixels that are equal to the no data value (default is 0) are interpreted as False, all other pixels as True.Multiband rasters are first evaluated band wise. The final mask for a given pixel is True, if all band wise masks for that pixel are True.

Output ClassProbabilitySample

Specify output path for sample (.pkl).


RegressionSample from Raster and Regression

Derives Regression sample from raster and regression.

Raster

Specify input raster.

Regression

Specify input raster.

Mask

Specified vector or raster is interpreted as a boolean mask.

In case of a vector, all pixels covered by features are interpreted as True, all other pixels as False.

In case of a raster, all pixels that are equal to the no data value (default is 0) are interpreted as False, all other pixels as True.Multiband rasters are first evaluated band wise. The final mask for a given pixel is True, if all band wise masks for that pixel are True.

Output RegressionSample

Specify output path for sample (.pkl).


UnsupervisedSample from ENVI Spectral Library

Derives unsupervised sample from ENVI spectral library.

ENVI Spectral Library

Select path to an ENVI (e.g. .sli or .esl).

Output Sample

Specify output path for sample (.pkl).


UnsupervisedSample from raster and mask

Derives unsupervised sample from raster and mask.

Raster

Specify input raster.

Mask

Specified vector or raster is interpreted as a boolean mask.

In case of a vector, all pixels covered by features are interpreted as True, all other pixels as False.

In case of a raster, all pixels that are equal to the no data value (default is 0) are interpreted as False, all other pixels as True.Multiband rasters are first evaluated band wise. The final mask for a given pixel is True, if all band wise masks for that pixel are True.

Output Sample

Specify output path for sample (.pkl).


Masking

Build Mask from Raster

Builds a mask from a raster based on user defined values and value ranges.

Raster

Specify input raster.

Foreground values

List of values that are mapped to True, e.g. [1, 2, 5].

Foreground ranges

List of [min, max] ranges, e.g. [[1, 3], [5, 7]]. Values inside those ranges are mapped to True.

Background values

List of values that are mapped to False, e.g. [1, 2, 5].

Background ranges

List of [min, max] ranges, e.g. [[-999, 0], [10, 255]]. Values inside those ranges are mapped to False.

Output Mask

Specify output path for mask raster.


Apply Mask to Raster. Pixels that are masked out are set to the raster no data value.

Raster

Specify input raster.

Mask

Specified vector or raster is interpreted as a boolean mask.

In case of a vector, all pixels covered by features are interpreted as True, all other pixels as False.

In case of a raster, all pixels that are equal to the no data value (default is 0) are interpreted as False, all other pixels as True.Multiband rasters are first evaluated band wise. The final mask for a given pixel is True, if all band wise masks for that pixel are True.

Masked Raster

Specify output path for raster.


Post-Processing

ClassProbability as RGB Raster

Creates a RGB representation from given class probabilities. The RGB color of a specific pixel is the weighted mean value of the original class colors, where the weights are given by the corresponding class propability.

ClassProbability

Specify input raster.

Output Raster

Specify output path for raster.


Random

Random Points from Classification

Randomly samples a user defined amount of points/pixels from a classification raster and returns them as a vector dataset.

Classification

Specify input raster.

Number of Points per Class

Has to be a number or a list of numbers. When a single integer number is given (e.g. 100), equalised random sample will be taken, i.e. in this case 100 samples per class. For taking a disproportional random sample, where the amount of samples should differ between classes, provide a list of numbers. This list has to have as many arguments as classes in the classification and has to be ordered according to the classes (e.g. ‘[100, 70, 90]’ in the case of three classes, 100 samples will be taken from the first class, 70 from the second, etc.). For a proportional stratified random sampling provide a float value between 0 and 1 (e.g. 0.3 for randomly drawing 30% of pixels in each class).

Output Vector

Specify output path for the vector.


Random Points from Mask

Randomly draws defined number of points from Mask and returns them as vector dataset.

Mask

Specified vector or raster is interpreted as a boolean mask.

In case of a vector, all pixels covered by features are interpreted as True, all other pixels as False.

In case of a raster, all pixels that are equal to the no data value (default is 0) are interpreted as False, all other pixels as True.Multiband rasters are first evaluated band wise. The final mask for a given pixel is True, if all band wise masks for that pixel are True.

Number of Points

Number of points to sample from mask.

Output Vector

Specify output path for the vector.


Regression

Fit GaussianProcessRegressor

Fits Gaussian Process Regression. See Gaussian Processes for further information.

RegressionSample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. See GaussianProcessRegressor for information on different parameters.

Output Regressor

Specifiy output path for the regressor (.pkl). This file can be used for applying the regressor to an image using ‘Regression -> Predict Regression’.


Fit KernelRidge

Fits a KernelRidge Regression. Click here for additional information.

RegressionSample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. See KernelRidge for information on different parameters. See GridSearchCV for information on grid search and StandardScaler for scaling.

Output Regressor

Specifiy output path for the regressor (.pkl). This file can be used for applying the regressor to an image using ‘Regression -> Predict Regression’.


Fit LinearSVR

Fits a Linear Support Vector Regression.

RegressionSample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. See LinearSVR for information on different parameters. See GridSearchCV for information on grid search and StandardScaler for scaling.

Output Regressor

Specifiy output path for the regressor (.pkl). This file can be used for applying the regressor to an image using ‘Regression -> Predict Regression’.


Fit RandomForestRegressor

Fits a Random Forest Regression.

RegressionSample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. See RandomForestRegressor for information on different parameters.

Output Regressor

Specifiy output path for the regressor (.pkl). This file can be used for applying the regressor to an image using ‘Regression -> Predict Regression’.


Fit SVR

Fits a Support Vector Regression.

RegressionSample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. See SVR for information on different parameters. See GridSearchCV for information on grid search and StandardScaler for scaling.

Output Regressor

Specifiy output path for the regressor (.pkl). This file can be used for applying the regressor to an image using ‘Regression -> Predict Regression’.


Predict Regression

Applies a regressor to an raster.

Raster

Select raster file which should be regressed.

Mask

Specified vector or raster is interpreted as a boolean mask.

In case of a vector, all pixels covered by features are interpreted as True, all other pixels as False.

In case of a raster, all pixels that are equal to the no data value (default is 0) are interpreted as False, all other pixels as True.Multiband rasters are first evaluated band wise. The final mask for a given pixel is True, if all band wise masks for that pixel are True.

Regressor

Select path to a regressor file (.pkl).

Output Regression

Specify output path for regression raster.


Transformation

Fit FactorAnalysis

Fits a Factor Analysis.

Sample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. See FactorAnalysis for information on different parameters.

Output Transformer

Specifiy output path for the transformer (.pkl). This file can be used for applying the transformer to an image using ‘Transformation -> Transform Raster’ and ‘Transformation -> InverseTransform Raster’.


Fit FastICA

Fits a FastICA (Independent Component Analysis).

Sample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. See FastICA for information on different parameters.

Output Transformer

Specifiy output path for the transformer (.pkl). This file can be used for applying the transformer to an image using ‘Transformation -> Transform Raster’ and ‘Transformation -> InverseTransform Raster’.


Fit FeatureAgglomeration

Fits a Feature Agglomeration.

Sample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. See FeatureAgglomeration for information on different parameters.

Output Transformer

Specifiy output path for the transformer (.pkl). This file can be used for applying the transformer to an image using ‘Transformation -> Transform Raster’ and ‘Transformation -> InverseTransform Raster’.


Fit Imputer

Fits an Imputer (Imputation transformer for completing missing values).

Sample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. See Imputer for information on different parameters.

Output Transformer

Specifiy output path for the transformer (.pkl). This file can be used for applying the transformer to an image using ‘Transformation -> Transform Raster’ and ‘Transformation -> InverseTransform Raster’.


Fit KernelPCA

Fits a Kernel PCA (Principal Component Analysis).

Sample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. See KernelPCA for information on different parameters.

Output Transformer

Specifiy output path for the transformer (.pkl). This file can be used for applying the transformer to an image using ‘Transformation -> Transform Raster’ and ‘Transformation -> InverseTransform Raster’.


Fit MaxAbsScaler

Fits a MaxAbsScaler (scale each feature by its maximum absolute value). See also examples for different scaling methods.

Sample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. See MaxAbsScaler for information on different parameters.

Output Transformer

Specifiy output path for the transformer (.pkl). This file can be used for applying the transformer to an image using ‘Transformation -> Transform Raster’ and ‘Transformation -> InverseTransform Raster’.


Fit MinMaxScaler

Fits a MinMaxScaler (transforms features by scaling each feature to a given range). See also examples for different scaling methods.

Sample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. See MinMaxScaler for information on different parameters.

Output Transformer

Specifiy output path for the transformer (.pkl). This file can be used for applying the transformer to an image using ‘Transformation -> Transform Raster’ and ‘Transformation -> InverseTransform Raster’.


Fit Normalizer

Fits a Normalizer (normalizes samples individually to unit norm). See also examples for different scaling methods.

Sample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. See Normalizer for information on different parameters.

Output Transformer

Specifiy output path for the transformer (.pkl). This file can be used for applying the transformer to an image using ‘Transformation -> Transform Raster’ and ‘Transformation -> InverseTransform Raster’.


Fit PCA

Fits a PCA (Principal Component Analysis).

Sample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. See PCA for information on different parameters.

Output Transformer

Specifiy output path for the transformer (.pkl). This file can be used for applying the transformer to an image using ‘Transformation -> Transform Raster’ and ‘Transformation -> InverseTransform Raster’.


Fit QuantileTransformer

Fits a Quantile Transformer (transforms features using quantiles information). See also examples for different scaling methods

Sample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. See quantile_transform for information on different parameters.

Output Transformer

Specifiy output path for the transformer (.pkl). This file can be used for applying the transformer to an image using ‘Transformation -> Transform Raster’ and ‘Transformation -> InverseTransform Raster’.


Fit RobustScaler

Fits a Robust Scaler (scales features using statistics that are robust to outliers). Click here for example. See also examples for different scaling methods.

Sample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. See RobustScaler for information on different parameters.

Output Transformer

Specifiy output path for the transformer (.pkl). This file can be used for applying the transformer to an image using ‘Transformation -> Transform Raster’ and ‘Transformation -> InverseTransform Raster’.


Fit StandardScaler

Fits a Standard Scaler (standardizes features by removing the mean and scaling to unit variance). See also examples for different scaling methods.

Sample

Specify path to sample file (.pkl).

Code

Scikit-learn python code. See StandardScaler for information on different parameters.

Output Transformer

Specifiy output path for the transformer (.pkl). This file can be used for applying the transformer to an image using ‘Transformation -> Transform Raster’ and ‘Transformation -> InverseTransform Raster’.


Transform Raster

Applies a transformer to an raster.

Raster

Select raster file which should be regressed.

Mask

Specified vector or raster is interpreted as a boolean mask.

In case of a vector, all pixels covered by features are interpreted as True, all other pixels as False.

In case of a raster, all pixels that are equal to the no data value (default is 0) are interpreted as False, all other pixels as True.Multiband rasters are first evaluated band wise. The final mask for a given pixel is True, if all band wise masks for that pixel are True.

Transformer

Select path to a transformer file (.pkl).

Transformation

Specify output path for raster.


InverseTransform Raster

Performs an inverse transformation on an previously transformed raster (i.e. output of ‘Transformation -> Transform Raster’). Works only for transformers that have an ‘inverse_transform(X)’ method. See scikit-learn documentations.

Raster

Specify input raster.

Mask

Specified vector or raster is interpreted as a boolean mask.

In case of a vector, all pixels covered by features are interpreted as True, all other pixels as False.

In case of a raster, all pixels that are equal to the no data value (default is 0) are interpreted as False, all other pixels as True.Multiband rasters are first evaluated band wise. The final mask for a given pixel is True, if all band wise masks for that pixel are True.

Transformer

Select path to a transformer file (.pkl).

Inverse Transformation

Specify output path for raster.