Self Organizing Map Based Radar Shadow Detection Using Textural Features

Inhalt:
This paper demonstrates the adaptation of a neural network based classification method to radar shadow detection process. The method which is previously used for automatic cloud detection and classification for satellite images using textural features is newly applied to radar shadow detection process. A novel utilization of Gray Level Co-occurrence Matrices (GLCM) method is preferred to extract textural features and Self Organized Maps (SOM) are used to classify the data according to pre-classified training set.