This paper presents a case study on environmental aspects related to the occurrence of visceral leishmaniasis in Teresina, Piauí, Brazil, from 1993 to 1996, in order to discuss the use of some appropriate geo-processing methods for median-resolution remote sensing images potentially useful for studying vector-borne transmissible diseases in urban areas. We present the main techniques: registration, geometric correction, restoration, fusion, segmentation, and classification. Using intra-class correlation indices applied to the proportion of area by class in the census tract, we compare four classifiers: Maxver, Bhattacharya, K-means, and Isoseg. This comparison was not devised to choose the best classifier, but to depict different classification scenarios aimed at recognizing the best identifiable image classes in urban settings. We conclude that even with limited resources, using low-cost and easily available median resolution images and free software to process and integrate information, it is possible to identify land use characteristics, potentially appropriate for analyzing urban areas exposed to environmental risk for vector-borne diseases.
Remote Sensors; Visceral Leishmaniasis; Endemic Diseases
