Per Block Urban Land Use Interpretation Using Optical Vhr Data and the Knowledge-based System Interimage

This study uses the InterIMAGE system and imagery from the QuickBird II optical sensor for the land use classification at two testsites in the metropolis of São Paulo, Brazil. InterIMAGE is an open source and free access system for knowledge-based image classification. Within InterIMAGE human knowledge is represented as a semantic net and by user-defined rules that can emulate human reasoning based on the paradigms of the object-oriented image analysis. The two land use classifications were carried out considering the urban blocks of the test-sites as analysis units. After obtaining the land cover classifications of the two test-sites, customized features related to the composition, structure and geometrical properties of the land cover objects within these blocks were created. The description of the land use classes was done associating these customized features with membership functions and arranging these expressions in membership values aggregation structures. The proposed methodology has shown to be promising for automatic land use interpretation of complex urban areas such as that found in São Paulo. The land use classifications achieved Overall Accuracies above 70% in the two test-sites and Kappa Indexes of 0.69 and 0.53. This study has explored some of the main presently available potentialities of InterIMAGE for object-based and knowledge-based image classification. This application of InterIMAGE, as well as the system itself, is under development.