An Exercise of Automatic Identification of Changes in Building's Representation in Cartographic Basis of Urban Areas

This paper presents the results of a digital aerophotogrametric image processing and LIDAR data for automatic detection of outdated information layers corresponding to buildings in cartographic databases of large-scale urban areas. The proposed technique consists of classical image processing routines applied to a large number of small images. These small images are clippings to the limits of the urban lots, automatically obtained a orthoretificated mosaic. The database used consists of: (a) a digital cartographic database with information layers associated with the boundaries of lots and projection of buildings on the land; (b) a georeferenced and orthoretificated aerophotogrammetric mosaic and (c) a cloud of laser scanner points of the study area. The operational procedure consists of the automatic trimming of the area associated with each lot in the mosaic, followed by the segmentation and classification of this image with altimetry data support to the laser cloud points and followed by the formation of geometries probable to be associated with the existing buildings. These procedures are performed without any user intervention, by a program implemented in a structured language whose input files and output data are in standard formats. A prototype system was implemented and tested in a database courtesy of Federal Patrimony Bureau, in the coastline of the state of Rio Grande do Sul - Brazil.