Producing a Building Change Map for Urban Management Purposes

Land information is required for the major activities of the municipal authorities. At this level, decision-making is supported by large-scale maps that include both topographic and thematic information. For these purposes, the efficient use of very high spatial resolution (VHR) imagery suggests the development of approaches that enable a timely discrimination, classification and delineation of urban elements according to quality standards. The current framework of the Portuguese large scale map production is very time-consuming and expensive, since it is based on manual editing of ortho-photographs to comply with very demanding technical specifications. Consequently, the time lag between availability of updated maps is generally 10 years. The main advantage of having a change map based on remote sensing data and a semi-automatic extraction methodology is to allow a more efficient mapping process, concentrating the updating efforts only in those areas or elements that changed. With such application cartographic products could be updated on an annual basis.

The work presented in this paper is the exploration of VHR imagery as an alternative source of geospatial information for large scale mapping to assist municipal urban planning in Portugal. The data base concerns spectral and altimetric data. For the spectral data base, multispectral and panchromatic QuickBird images from 2005, of the city of Lisbon, are available. Furthermore, an altimetric data set composed by a LiDAR (Light Detection And Ranging) image of the second pulse from 2006 and a Digital Terrain Model (DTM) are also available. The methodology was tested over a study area of 64 ha, located in Lisbon. This area is characterized by several building typologies that include industrial properties, schools, apartments and single-family housing.

The QuickBird images were orthorectified and subjected to a pansharp fusion. The LiDAR data and the DTM were used to produce the normalized Digital Surface Model (nDSM). The extraction methodology was applied to this data set using a feature extraction software, in order to produce a map of the buildings present in the image. After building extraction, a post-processing was conducted to enhance the geometric quality of the elements. Afterwards, a quality assessment was performed. The assessment was exhaustive and involved comparisons of extracted features against a reference dataset collected by visual interpretation of the imagery.

Having the 2005 building map, the next step was to produce a changed map using the municipal map, at 1:1000 scale, from 1998. The change detection process was able to identify missing structures and to detect new ones. However, its geometric and thematic quality is not yet sufficient to allow a direct updating of the municipal databases. The goal is rather to produce an alarm system that indicates the location of potential changes in the building areas. This layer can be used by the municipal technicians as the basis for manual editing, following the technical specifications indicated for the 1:1 000 scale.