Information Mining via Coherence Estimation from Multilook Incoherent SAR Imagery

In this work, an application of "browsing by coherence" is presented. An interferometric pair of single-look complex (SLC) ERS-1/2 Tandem images is multi-look processed and the resulting real-valued images are further compressed through a hierarchical method to yield quick-looks that may be browsed at different spatial resolutions, i.e. number of looks. Pairs of quick-look icons are utilized to estimate coherence by means of a new method based on measurements of temporal correlation of speckle. The outcome maps are compared with spatially degraded versions of the coherence map calculated from the original SLC data. Experiments are aimed at showing that the method yields steady results varying with the number of looks of icons that are browsed.