Data fusion approach for Urban area identification using multisensor information

This article presents a procedure to identify and extract urban areas in medium resolution satellite images. At present, we have and continue to study various methodologies to process and extract information on urban surfaces, since urban growth is having environmental impacts on the involved ecological systems. The proposed method takes advantage of the fact that data fusion allows us to combine in an optimal manner, multiple sources of classifiers and to generate a single source of information. In this context, we propose the use of data fusion algorithms, by multiple classifiers, taking into account the spectral and spatial characteristics of the satellite data, which in our case are the Landsat ETM+ and the ENVISAT-ASAR. The developed system includes an ensemble fusion architecture and the use of algorithms such as Fuzzy K-mean and Markov Random Field (MRF). The study case is the Guadalajara metropolitan area, in Jalisco, Mexico, which has great growth and sprawl; in its surrounding areas there are regions which are interesting in terms of geothermal exploitation and with great ecological value. The experimental results, using the multiple classifier system (MCS) show the urban characteristics at the regional scale, offering results that are potentially significant at this scale and the direction of changes in urban growth.