Novel Seed Point Selection Technique for Dual-Pol SAR Images

Abstract

In recent years, dual polarimetry has gained much importance among other earth observation missions due to its less complex architecture, larger swath coverage and low battery consumption. In this paper, a novel technique for automatic initial seed point selection for dual-pol data is proposed and its performance for different dual-pol combinations has been assessed. The proposed technique is independent of polarization combinations. Impact on classification when only 50% pixel is considered for selecting initial seed point is demonstrated. RISAT-1 quad-polarimetric data has been used for validating our result.