Unsupervised image segmentation based on a self-organizing feature map and a texture measure
Ari J. E. Visa · 2003
A new approach to unsupervised texture segmentation is represented. The method is based on a local texture measure, a grey tone spatial dependence matrix. The randomly sampled local measures self-organize to a topological feature map. The topological feature map is used as a set of reference vectors later on when the whole image is processed in raster scan manner by the local texture measurement. The label of a region is the address on the topological feature map. The interpretation of a label is given by identified samples. The method has been applied in segmentation of remote sensing images and aerial photographs.>