Image analysis methodologies applied to recognition of cartographic features

One of the many applications of mathematical morphology (MM) is cartographic updating. The automatization of this task is culturally interesting for the importance of an awareness of the actual state of the Earth's surface, it is scientifically challenging and it has a strong economic interest. The actual technology has enormous computing possibilities that can be focused on this subject through MM The aim of this work is to apply this mathematical theory in order to obtain a method to recognize automatically forms (segmentation) in satellite images, proposing a concept and a sequence of morphological operators based on data belonging both to the spatial domain and to the spectral domain of the treated images. The main item of this thesis is a variant of the concept of the mosaic image and transformations relying on it. These are shown and compared to others in a case study using satellite images of the city of Luanda, in Angola, but they can be easily adapted to any other digital image.