Análise multi-temporal de cenas do satélite quickbird usando um novo paradigma de classificação de imagens e inferências espaciais. Estudo de caso: Belo Horizonte, MG / Multi-temporal analysis of quickbird images using a new classification paradigm and spatial inferences. A case study: Belo Horizonte - MG, Brazil

AUTOR(ES)
DATA DE PUBLICAÇÃO

2006

RESUMO

Due to the strong urbanization process occurring in Brazil and of population increase in cities, and consequently to serious impacts on environment, there is a growing need to develop methodologies and technologies which allow to monitor, analyze and to make an intervention at the urban expansion. Belo Horizonte (State Capital of Minas Gerais) is a highly complex metropolitan area, making a diagnosis of its problems very difficult. Using high resolution images, it is possible to obtain information needed for planning activities, in order to improve quality of life. Applying automatic land use classification algorithms which use new paradigms, relevant information can be extracted to improve understanding of the city. Object-oriented classification which uses context relations, hierarchy and fuzzy logic, are new and efficient procedures to map urban land cover, specially when working with high-resolution images, like those from QUICKBIRD satellite and ancillary data. These datasets and multi-temporal analysis allow spatial inferences, generating new knowledge on urban areas. Meanwhile, to perform multi-temporal studies with high-resolution satellite images, all data must be duly co-registered and, in order to do that, geometric correction models must be applied. The distortions suffered by these images impede its direct superposition on thematic cartographic data. In this study, spatial inferences were made from two quarters of Belo Horizonte, based on QUICKBIRD images from 2002 and 2004, using object-oriented classification as well as geological, geotechnical and legal data. An ortho-rectification of both datasets was done, considering the rigorous model, allowing data superposition in a GIS. So it is possible to generate information on the expansion of these quarters in the timeframe 2002-04, the illegal land occupation and areas with landslide / erosion risks which could cause damage to property and individuals.

ASSUNTO(S)

planejamento urbano image analysis image classification classificação de imagens satellite imagery high resolution alta resolução geometric rectification belo horizonte (mg) correção geométrica análise de imagens urban planning sensoriamento remoto imagens de satélite remote sensing

ACESSO AO ARTIGO

Documentos Relacionados