Object Segmentation in Elevation Space Using Mathematic Morphology

摘要

The original Digital Surface Model (DSM) produced from image matching in digital Photogrammetry or from scanning by Airborne Laser Scanner (ALS) are the mixing representation for ground, buildings, trees and many others. The ground surface and those objects above the ground may be segmented from the DSM. We propose two mathematic morphology methods, Top-Hat Transformation and H-Dome Transformation, to reconstruct DSM to separate those objects from the ground surface. The procedure of the proposed methods includes: (1)Top-Hat Transformation to filter out the noise in elevation space. (2)H-Dome Transformation to detect the local regional maxima as the objects above the ground. Experimental results indicate that the proposed method might reach high reliability. It is also found that those objects above the ground may further be used in building detection provided that other inputs are available, and DTM production.