POLSAR data classification using Self-Organizing Map

In this paper, we present a supervised classification method based on the SOM algorithm to classify POLSAR data. Three parameters of Krogager decomposition were applied as input features of the SOM algorithm. At first the SOM was training and a cluster map was generated by the weight vectors of the SOM’s nodes. Secondly ground truth data are reinputted into the SOM to find the label of each cluster and land cover map was produced. Test image acquired by AIRSAR/JPL in L-band was used to demonstrate the advantage of this algorithm. We compare the SOM result with K-Means algorithm's result