Laser scanning data segmentation in urban areas by a Bayesian framework
Galvanin, Edineia Aparecida dos Santos ... [et al.]
In this paper is presented a region-based methodology for segmentation of Digital Elevation Model (DEM) obtained from laser scanning data. The methodology is based on two sequential techniques, i.e., a recursive splitting technique using the quad tree structure followed by a region merging technique using the Markov Random Field (MRF) model (Conditional Autoregressive model CAR). The recursive splitting technique starts splitting the DEM into homogeneous regions. However, due to slight height differences in DEM, region fragmentation can be relatively high. In order to minimize the fragmentation, a region merging technique based on the Bayesian framework is applied to the previously segmented data. The resulting regions are firstly structured by using the neighborhood structure. Thus, two regions have connectivity between them if corresponding regions share a common boundary. Next it is used a hierarchical model, whose height values in the data depend on a general mean plus some random effect. Following the Bayesian paradigm, it was consider for the random effects a CAR prior. The posterior probability distribution was obtained by Gibbs sampler. Regions presenting high probability of similarity are merged. Experiments carried out with laser scanning data DEM showed that the methodology allows to obtain the objects in the DEM with a low amount of fragmentation.
Event: XXIII International FIG Congress : Shaping the change
Only personal, non-commercial use of this document is allowed.