On the ensemble of multiscale object-based classifiers for aerial images: a comparative study

Autor(es)

Esmael, Agnaldo Aparecido ^[1] ; dos Santos, Jefersson Alex ^[2] ; Torres, Ricardo da Silva ^[1]

Afiliação do(s) autor(es)

^[1] Univ Estadual Campinas, Inst Comp, Av Albert Einstein 1251, BR-13083852 Campinas, SP - Brazil

^[2] Univ Fed Minas Gerais, Dept Comp Sci, Av Antonio Carlos 6627 Predio ICEx, BR-31270010 Pampulha Belo Horizonte, MG - Brazil

• Tipo de documento: Artigo Científico
• Fonte: MULTIMEDIA TOOLS AND APPLICATIONS; v. 77, n. 19, p. 24565-24592, OCT 2018.
• Citações Web of Science: 0

Resumo

Remote sensing images (RSIs) are increasingly used as data source to produce maps used in several applications. Modern sensors launched into space from the end of the 1990s have been producing high spatial resolution RSIs. The use of classification methods based on regions, called as Geographic Object-Based Image Analysis (GEOBIA), has been demonstrated to be more appropriate to deal with this kind of image. However, finding the appropriate segmentation scale, which is not a trivial task, is crucial for the success of a GEOBIA method. In this paper, we perform a comparative study involving seven methods for RSI multiclass classification that combine different features extracted from different scales: M1-OvA, M2-OvO, M3-AdaMH, M4-Samme, M5-MV, M5-WMV, and M6-Cascade. The first four methods are boosting-based techniques and the last three are based on the majority vote approach. The effectiveness of the proposed methods was evaluated by analyzing the results of experiments conducted in three RSIs datasets. The methods were compared with the baseline SVM with Kernel RBF by measuring the overall accuracy, the Kappa Index, and the accuracy per class. The results show that all the proposed methods are effective for RSI classification. (AU)

• Processo FAPESP: 14/50715-9 - Characterizing and predicting biomass production in sugarcane and eucalyptus plantations in Brazil
• Beneficiário: Rubens Augusto Camargo Lamparelli
• Linha de fomento: Auxílio à Pesquisa - Parceria para Inovação Tecnológica - PITE
• Processo FAPESP: 16/50250-1 - O segredo de jogar futebol: Brasil versus Holanda
• Beneficiário: Claudio Alexandre Gobatto
• Linha de fomento: Auxílio à Pesquisa - Temático
• Processo FAPESP: 17/20945-0 - EMU concedido no processo 16/50250-1: local positioning system
• Beneficiário: Ricardo da Silva Torres
• Linha de fomento: Auxílio à Pesquisa - Programa Equipamentos Multiusuários
• Processo FAPESP: 14/12236-1 - AnImaLS: Anotação de Imagem em Larga Escala: o que máquinas e especialistas podem aprender interagindo?
• Beneficiário: Alexandre Xavier Falcão
• Linha de fomento: Auxílio à Pesquisa - Temático
• Processo FAPESP: 13/50169-1 - Towards an understanding of tipping points within tropical South American biomes
• Beneficiário: Ricardo da Silva Torres
• Linha de fomento: Auxílio à Pesquisa - Parceria para Inovação Tecnológica - PITE
• Processo FAPESP: 13/50155-0 - Combining new technologies to monitor phenology from leaves to ecosystems
• Beneficiário: Leonor Patricia Cerdeira Morellato
• Linha de fomento: Auxílio à Pesquisa - Programa de Pesquisa sobre Mudanças Climáticas Globais - PITE
• Processo FAPESP: 15/24494-8 - Comunicação e processamento de big data em nuvens e névoas computacionais
• Beneficiário: Nelson Luis Saldanha da Fonseca
• Linha de fomento: Auxílio à Pesquisa - Temático