Mapping land use cover using segmentation and classification of fraction images, soil, shade and vegetation, derived from a linear mixing model applied to Landsat-5 TM data, Tucuruí reservoir region - Pará

Remote sensing techniques are mandatory for monitoring land use changes in large areas such as the Amazon. Land use mapping is usually performed by both manual and digital pixel based classification methods which are cost and time-consuming. In this study an image segmentation approach is applied to unmix TM-Landsat images for mapping land use classes in Tucuruí-PA reservoir region to 1996 and 2001.

mapping the land use; remote sensing; image segmentation