Major Update: Semi-Automatic Classification Plugin v. 6.4.0

A new update of the Semi-Automatic Classification Plugin v. 6.4.0 has been released which includes a few new features and improvements.
I'm also very happy that the Semi-Automatic Classification Plugin is now included in the featured QGIS plugins. Many thanks to the QGIS community and to Paolo Cavallini, chair of the QGIS project group.

Following the changelog:

-in Multiple ROI creation tool added a stratification option based on raster values
-in Accuracy tool added the calculation of area based error matrix
-optimization of raster computation for several functions such as DOS1 and accuracy and classification report
-fixed symbology of classification raster
-bugfixing

This updates add two features that are useful for accuracy assessment (a new tutorial about this will be released soon), which are the stratification for the random sample selection (using the tool Multiple ROI Creation), and the calculation of area based error matrix (using the tool Accuracy).

The Multiple ROI Creation option "Stratified for the values" allows for the definition of one or more expressions for creating random points inside defined raster values. If checked, it creates random points inside the values defined in the expression calculated for the first band of the defined band set; the expression must include the variable raster (e.g. raster > 1) or multiple expressions separated by semicolon (e.g. raster == 1; raster > 5).

This allows for the creation of stratified random samples that can be used for accuracy assessment.

The Accuracy tool now calculates the area based error matrix (according to “Olofsson, et al., 2014. Good practices for estimating area and assessing accuracy of land change. Remote Sensing of Environment, 148, 42 – 57”) where each element represents the estimated area proportion of each land cover class. This allows for estimating the unbiased user’s accuracy and producer’s accuracy, the unbiased area of classes according to reference data, and the standard error of area estimates and the confidence intervals. Also, watch this useful webinar about unbiased area estimation by NASA ARSET. 

Also, the computational speed of several functions have been improved such as DOS1, accuracy and classification report.

Finally, the symbology of classification rasters have been fixed, and now it can be modified in QGIS symbology.

For any comment or question, join the Facebook group about the Semi-Automatic Classification Plugin.