Abstract
The coherent nature of SAR processing enables the generation of interferometric image pairs, in which the measure of coherence contains information about the temporal stability of the scene. This feature of repeat-pass interferometric data can be used to perform coherent change detection. Aiming towards this application, the contrast or rather the gray level differences between coherent and incoherent regions need to be maximized, so that subsequent change detection steps can perform best. Inspired by the classical coherence estimate, in this paper a workflow is presented to pre-process the image data for coherent change detection. The workflow contains several filtering steps to improve the contrast between coherent and incoherent areas of an image. It consists of three steps: Speckle filtering of the amplitudes, coherence-assisted removal of the topographic phase, and coherence-guided noise-filtering of the interferometric phase. The approach was tested with several speckle filters and several coherence estimation schemes.