Serviços Personalizados
Journal
Artigo
Indicadores
Links relacionados
- Similares em SciELO
Compartilhar
Investigaciones geográficas
versão On-line ISSN 2448-7279versão impressa ISSN 0188-4611
Resumo
GARCIA MORA, Tzitziki Janik e MAS, Jean-François. Comparison of methodologies for mapping land use cover in Southeast Mexico. Invest. Geog [online]. 2008, n.67, pp.7-19. ISSN 2448-7279.
Digital classification of satellite images has become an important tool for land use cover monitoring. There have been many studies and researches aiming to develop and assess classification methods which are capable of producing results with higher accuracies in order to support decision making. Through a literature review, we found various classification methods that have been tested and are able to produce better results than the conventional one. This work compares different classification methods in a southeast region of Mexico. Results show that with ancillary data which were represented by prior probabilities, the classifications obtained 20% more accuracy that those by conventional classification methods, and even better than those methods using non-parametric classifier and contextual information.
Palavras-chave : Contextual classification; neural networks; ancillary data.