Importance of speckle filtering in image classification of SAR data

In this work an attempt is made to study the impact and importance of speckle filtering for classification using Advanced Land Observing Satellite Phase Array L band Synthetic Aperture Radar (ALOS PALSAR-2) data. It provides an opportunity to study vegetation, as it uses L band frequency, therefore, penetration capability is more and helps us to study the deforested area. Amazon Forest, Roraima State in Brazil, South America has been chosen as study area. Speckle is a general phenomenon in SAR imagery caused by the interaction of the out-of-phase waves reflected from a target. Presence of speckle in SAR images degrades the interpretability of the land features in the data. We presented a comparative study of several speckle filters that are widely used in radar imaging community. Speckle evaluation is done by Speckle Suppression Index (SSI). Based on the SSI values for a filter, less speckle image is opted for image classification. K Mean clustering technique is used for image classification.