Elaboration of a Hybrid Method for the Enhancement of Buildings Reconstruction in Multispectral Images

Abstract

Remote sensing became a current research area, therefore many detection methods of urban areas were elaborated, some of them are promising, others still less efficient, For this purpose we propose in this article an enhancement of methods that we have already elaborated previously based on new ideas and especially on the mathematical morphology. The method is inspired by the top hat (TH) to better detect dark structures, and to correct false detections a hit or miss transform (HMT) is applied, and for a better noise suppression we improved the sequential alternating filter by adding the median filter between his two steps. Before concluding, we present a comparative study with methods that we have improved before; to show the effectiveness of our method which gives us the best result for the entire used satellite images database, with a good detection rate which crossed 97%.