Major Update for the Semi-Automatic Classification Plugin for QGIS

This post is to inform you of a major update for the Semi-Automatic Classification Plugin for QGIS. Following the changes of the QGIS API, I have released two plugin versions:
  • version 1.8 is compatible with QGIS 1.8;
  • version 2.0 is compatible with QGIS 2.0.

Both versions have the same functionalities, and the code of the version 2.0 has been updated to the API changes of the new QGIS 2.0.
The major changes from the previous version 1.5.1 of the plugin are:
  • Improved interface layout;
  • The plugin can calculate the spectral signatures of ROIs;
  • A new "Spectral signature" tab allows for the visualization of the spectral signatures in a chart;
  • A new "Band set" tab allows for the definition of a group of single bands as image input.
Following, more details about these improvements.

Improved interface layout

The interface is now composed of 5 tabs:
  • "ROI tool", for the creation of ROIs;
  • "Classification", for the preview and classification of the image;
  • "Spectral signature", for the visualization of ROI spectral signatures in a chart;
  • "Band set", for the definition of a group of single bands as classification input;
  • "Settings", for the advanced settings.

Also, a progress bar is shown during the main processes; a "Cancel" button allows the user to stop the process, but the process is stopped after the completion of the very last Python command.
Semi-Automatic Classification Plugin interface

Spectral signatures

It is possible to calculate the spectral signature during the ROI definition.
ROI statistics (Mean and Standard Deviation for each band) are stored in the shapefile, which is created with the required fields according to the number of raster bands. Also, the plugin can add the required signature fields to any existing shapefile.
Attribute table of a shapefile with ROI statistics

Spectral signature tab

Band set definition

The Semi-Automatic Classification Plugin can work with multispectral images, also if the bands thereof are separated files (single band images). The "Band set" tab allows for the definition of a group of images (i.e. a band set) loaded in QGIS, which is used as input image for the ROI creation and the classification processes.
This is especially useful for hyperspectral images, because the classification process is rapider with separated bands. Band sets can be saved and imported, for the purpose of testing the classification results with different band compositions.
If a band set is loaded, the item << band set >> appears in the raster lists of the "ROI tool" and "Classification" tabs.

 The Semi-Automatic Classification Plugin will be soon available from the QGIS repository; if you want, you can download the 1.8 version from here, or the 2.0 version from here.
For more information about the interface see here.