Application of neural unsupervised methods to environmental factor analysis of multi-spectral images with texture features

Indicizzato in scopus con codice eid=2-s2.0-62949209534 In this paper, we present a Kohonen's Self Organizing Map for the land-cover classification of multi-spectral satellite images. In order to obtain an accurate segmentation we introduced as input for the network, in addition to the spectral data, some texture measures which gives a contribution to the classification of manmade structures. The texture features were extracted from high resolution images by means of Gray Level Co-occurrence Matrix (GLCM) and standard deviation. After clustering of SOM outcomes, we associated each cluster with a major land cover and compared them with prior knowledge of the scene analyzed. The results are encouraging as showed by the high values of the accuracy