Further results on Kalman filters for speckle noise reduction on SAR images

The modified adaptive block Kalman filter (MABKF) developed by Geling and Ionescu [1994] was designed to reduce the level of speckle in complex SAR images including urban regions where normal speckle assumptions are no longer true. In order to improve the performance of the filter two modifications have been developed. The MABKF filter methodology has been combined with a dynamic model using a symmetric full-plane region of support. Additionally, a multiplicative noise term has been added to the dynamic equation to model the highly variable nature of a complex urban scene containing many strong reflectors. The filters are demonstrated on an ERS-1 SAR image of Victoria, B.C.