Abstract:

Based on the analysis of occluded buildings in high resolution SAR images, a novel method for occluded building height estimation was proposed. First, a projection model for SAR building signature prediction from a 3D pa-rametric model was constructed. Second, a likelihood function for SAR building signature evaluation was formulated. Finally, the parametric models with different height parameters were evaluated with the actual SAR image by the likelih-ood function. The estimated building height corresponded to the parametric model which reached a maximum likelihood value. Comparative experimental results show that the proposed method is more effective in estimating building height even under partial occlusion.

Key words: HR SAR image, building height estimation, partial occlusion, SAR building signature, likelihood function