Rubble Detection from VHR Aerial Imagery Data Using Differential Morphological Profiles

Rubble Detection from VHR Aerial

 Imagery Data Using Differential

Morphological Profiles

G.K. Ouzounis*, P. Soille, M. Pesaresi

Geo-Spatial Information Analysis for Global Security and Stability, Global Security and Crisis Management Unit,

Joint Research Centre, European Commission, Via E.Fermi 2747, I-21027 Ispra (VA), Italy

(georgios.ouzounis, pierre.soille, martino.pesaresi)@jrc.ec.europa.eu 

34th International Symposium on 

Remote Sensing of Environment

The GEOSS Era: Towards Operational Environmental Monitoring

Sydney, Australia

10-15 April 2011

Abstract –

  Rubble detection is a key element in post disaster crisis assessment and response procedures. In this paper we present an automated method for rapid detection and quantification of rubble from very high resolution (VHR) aerial imagery of urban regions. It is a two step procedure in which the input image is projected onto a hierarchical representation structure for efficient mining and decomposition. Image features matching the geometric and chromatic properties of rubble are fused into a rubble layer that can be re-adjusted interactively. The targeted objects are evaluated based on a density metric given by spatial aggregation. The method is tested on a small-scale exercise on the publicly available aerial imagery of Port-au-Prince, Haiti. Performance and preliminary results are discussed. 

Keywords: Rubble detection, earthquake, differential area profile, spatial aggregation.

5. CONCLUSIONS 

  The proposed method was tested for the case of a simple dichotomy, i.e. detecting rubble against the global background. The detection success rate for the tile of Figure 3 was approximately 92%, suggesting that the method in its simplest form is sufficiently reliable for rapid damage assessment. In future work, we aim at utilizing the local background information layer to minimise the false alarms. Richer set of constraints, based on attribute vectors, are investigated for computing more delicate decompositions. Moreover, an automatic assessment method is being developed to compare the results of our method against the ground truth (JRC,2010) in a full scale exercise on the entire Portau- Prince dataset (360GB), currently under preparation

Full Text

Download                        Click here

Read and download       Click here