Object oriented image analysis of urban areas whith LANDSAT images

Abstract

This study attempts to develop an automatic approach to classify continuous and disperses urban areas that characterize the urban sprawl, in high urbanization regions of São Paulo State. The objective is to achieve a methodology for the increase of precision and reduction of processing time, to be applied in other regions of Brazil. Object Oriented Image Analysis with Definiens software and LANDSAT images of the São José dos Campos municipality of 1985 and 1999 were used. This procedure consist of multiresolution image segmentation and classification based on a logical fuzzy. Considering the resolution of the images
used, and the relief of the study area the Object Oriented Image Analysis have a good performance to map urban areas. The results obtained show that the developed methodology can be replicate in other urban areas.