Rank Aggregation for Pattern Classifier Selection in Remote Sensing Images
Nenhuma Miniatura disponível
Data
2014-04-01
Autores
Faria, Fabio A.
Pedronette, Daniel C. G. [UNESP]
Santos, Jefersson A. dos
Rocha, Anderson
Torres, Ricardo da S.
Título da Revista
ISSN da Revista
Título de Volume
Editor
Institute of Electrical and Electronics Engineers (IEEE)
Resumo
In the past few years, segmentation and classification techniques have become a cornerstone of many successful remote sensing algorithms aiming at delineating geographic target objects. One common strategy relies on using multiple complex features to guide the delineation process with the objective of gathering complementary information for improving classification results. However, a persistent problem in this approach is how to combine different and noncorrelated feature descriptors automatically. In this regard, one solution is to combine them through multiple classifier systems (MCSs) in which the diversity of simple/non-complex classifiers is an essential issue in the definition of appropriate strategies for classifier fusion. In this paper, we propose a novel strategy for selecting classifiers (whereby a classifier is taken as a pair of learning method plus image descriptor) to be combined in MCS. In the proposed solution, diversity measures are used to assess the degree of agreement/disagreement between pairs of classifiers and ranked lists are created to sort them according to their diversity score. Thereafter, the classifiers are also sorted according to their performance through different evaluation measures (e. g., kappa and tau indices). In the end, a rank aggregation method is proposed to select the most suitable classifiers based on both the diversity and the effectiveness performance of classifiers. The proposed fusion framework has targeted at coffee crop classification and urban recognition but it is general enough to be used in a variety of other pattern recognition problems. Experimental results demonstrate that the novel strategy yields good results when compared to several baselines while using fewer classifiers and being much more efficient.
Descrição
Palavras-chave
Coffee crop classification, diversity measures, information fusion, meta-learning, urban recognition
Como citar
Ieee Journal Of Selected Topics In Applied Earth Observations And Remote Sensing. Piscataway: Ieee-inst Electrical Electronics Engineers Inc, v. 7, n. 4, p. 1103-1115, 2014.