All the risks concerning the long lasting exposition to the asbestos, for instance diseases affecting the respiratory system, are known very well. In particular, it’s the friable asbestos the most dangerous element, above all as for the easily dispersing fibres. Despite the above mentioned risks, the number of asbestos covered block is still high in many nations, in particular they are buildings characterized by eternit coverings (a mixture of cement and asbestos), certainly due to the features of materials and their low cost.Various legislations implement reclamation works on the asbestos roofs, but in several countries building farms keep on using such material, despite it’s forbidden to them. The present work has as objective one to supply a cheap tool for the automatic identifying with the buildings which do use for its covering of this very dangerous technology; they are taken back three alternative methodologies of development of the tool, everyone working on remote sensed images and of a procedure of specific classification for the problems in exam, skilled actually turned out to offer useful on the applicatory plan. In the specific one the three configurations melt their work on the comparison between the spectrum of a building, spatially considered well known, characterized by covering in cement-asbestos with the spectrum of the pure material obtained from laboratory measures.From such comparison it is generated, in the first two cases, the building and the implementation of a suitable operator that through the estimate of an opportune equalization coefficient (built respectively using the neural nets and the ionized plasma (electromagnetic fields)), it allows to identify in automatic way the present coverings on the territory and therefore proceed to the next phase of pixel-oriented classification of the whole image.The last methodology is responsible for making a comparison between the classifications obtained with the methodologies preceding (pixel-oriented) and that coming from the use of an object-oriented method consisting in a Nearest Neighbor classification tool following a multi-resolution segmentation worked on the whole scene.

A Comparison Between Methods – A Specialized Operator, Object Oriented and Pixel Oriented Image Analysis – To Detect Absestos Coverages in Building Roofs Using Remotely Sensed Data

BARRILE, Vincenzo;
2008-01-01

Abstract

All the risks concerning the long lasting exposition to the asbestos, for instance diseases affecting the respiratory system, are known very well. In particular, it’s the friable asbestos the most dangerous element, above all as for the easily dispersing fibres. Despite the above mentioned risks, the number of asbestos covered block is still high in many nations, in particular they are buildings characterized by eternit coverings (a mixture of cement and asbestos), certainly due to the features of materials and their low cost.Various legislations implement reclamation works on the asbestos roofs, but in several countries building farms keep on using such material, despite it’s forbidden to them. The present work has as objective one to supply a cheap tool for the automatic identifying with the buildings which do use for its covering of this very dangerous technology; they are taken back three alternative methodologies of development of the tool, everyone working on remote sensed images and of a procedure of specific classification for the problems in exam, skilled actually turned out to offer useful on the applicatory plan. In the specific one the three configurations melt their work on the comparison between the spectrum of a building, spatially considered well known, characterized by covering in cement-asbestos with the spectrum of the pure material obtained from laboratory measures.From such comparison it is generated, in the first two cases, the building and the implementation of a suitable operator that through the estimate of an opportune equalization coefficient (built respectively using the neural nets and the ionized plasma (electromagnetic fields)), it allows to identify in automatic way the present coverings on the territory and therefore proceed to the next phase of pixel-oriented classification of the whole image.The last methodology is responsible for making a comparison between the classifications obtained with the methodologies preceding (pixel-oriented) and that coming from the use of an object-oriented method consisting in a Nearest Neighbor classification tool following a multi-resolution segmentation worked on the whole scene.
2008
Hazards Recognition, Feature Detection, Neural Networks, Multiresolution Segmentation
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.12318/288
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 7
  • ???jsp.display-item.citation.isi??? ND
social impact