Comparing Non-coherent and Coherent techniques for the Detection of Scene changes from multi-temporal SAR imagery

Abstract:
In this work several techniques, both coherent and non-coherent, for detecting scene changes from multitemporal SAR images are discussed and their performance evaluated and compared. Assuming given scenario conditions this paper aims at discussing and comparing the performance of the different approaches for both the non-adaptive (a priori knowledge of the scene correlation characteristics) and adaptive (estimated characteristics) cases. Moreover, considering the complementary information given by the non-coherent and coherent approaches, also a decentralized combination is studied. Reported results show that the adaptive change detection techniques show performance close to the corresponding non adaptive ones and that the joint use of non-coherent and coherent approaches can sensibly improve the performance with respect to those achievable by the single technique.