DSM GENERATION FROM STEREOSCOPIC IMAGERY FOR DAMAGE MAPPING, APPLICATION ON THE TOHOKU TSUNAMI
Résumé
Change detection methods from remote sensing are largely investigated, especially for damage mapping after a disaster such as tsunami, earthquake, flood or landslide. In this context, a fast evaluation and localization of the inferred damages is essential for the rescue teams and the authorities [1], [2], [3], an automatic method is then particularly adapted. Today, most of the automatic or semi-automatic change detection methods are based on radiometric changes [2] but a significant amount these changes are irrelevant. An alternative is a change detection method involving objects elevation [4]. The advantage of using this information is that most of the indexed changes are of interest, especially in an urban context. A previously submitted article describes the fully automatic and generic processing flow we developed for the production of change detection maps from stereoscopic images. This flow consists in the generation of accurate Digital Surface Model (DSM) with an improved image-space based matching technique followed by the detection of real changes from the DSM difference through a classification method with a spatial regularization constraint. In this article, we present the application of this methodology on a real application case, the 2011 earthquake and tsunami occurred on the Tohoku region, in Japan.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...