QGIS plugin for Image Analysis, featuring image segmentation, segment statistics calculation and segment label propagation.
This plugin was firstly developed as separate plugins as part of a Pattern Recognition class final project at the Applied Computation Graduation Program at the Brazilian National Institute for Space Research (INPE).
Just download the .zip and extracts the content on your qgis plugin folder (HOME/.qgis2/python/plugins on *NIX platforms)
Given a raster image, this method segments it by first clustering with a k-means method and then labeling connected components. Output is a shapefile with all the segments found.
After the raster image is segmented, this method computes mean, median and standard deviation for each segment, based on all the pixels of the original raster image that resides inside that segment. Output is saved on the .dbf file of the segmented shapefile.
Given a segments shapefile, and a ROIs shapefile, a SVM is trained with the intersection of the two shapefiles, which is then used to propagate the labels to the rest of the segments. Output is a new classified shapefile.
This plugin is released under the GNU GPL License.