%0 Conference Proceedings
%4 sid.inpe.br/marte2/2015/06.15.15.30.49
%2 sid.inpe.br/marte2/2015/06.15.15.30.50
%@isbn 978-85-17-0076-8
%F 729
%T Avaliação da redução do efeito de sombreamento em imagens multiespectrais de áreas urbanas utilizando componentes principais e análise de agrupamento
%D 2015
%A Azevedo, Samara Calçado de,
%A Tachibana, Vilma Mayumi,
%A Silva, Erivaldo Antônio da,
%@electronicmailaddress samara_calcado@hotmail.com
%E Gherardi, Douglas Francisco Marcolino,
%E Aragão, Luiz Eduardo Oliveira e Cruz de,
%B Simpósio Brasileiro de Sensoriamento Remoto, 17 (SBSR)
%C João Pessoa
%8 25-29 abr. 2015
%I Instituto Nacional de Pesquisas Espaciais (INPE)
%J São José dos Campos
%P 3707-3714
%S Anais
%1 Instituto Nacional de Pesquisas Espaciais (INPE)
%X Shadows represent a problem for feature extraction in urban areas, mainly when dealing with high spatial resolution remote sensing images. This work aims to evaluate one proposed method to remove shadow effects in multispectral images. Two approaches have been followed and compared. First, each band was corrected individually and principal components analysis (PCA) used to reduce high correlation that exists among images and the quantity to be processed. In the second approach, the PCA was firstly applied and corrected after, in order to verify the correction behavior in transformed data. To alleviate the shadow effects, priori detection of shadow areas using black top-hat morphological operator and automatic binarization were done, and then, these areas were corrected by local histogram matching from neighborhoods statistics. The method was tested in WorldView-2 multispectral images from São Paulo city, which is one of the densest urban areas in the world. Experimental results and visual analysis have shown a decrease in radiometric differences caused by shadows in both approaches. Also, an increase on distinction between features that were occluded was observed and proven by unsupervised classification using k-media. However, further analysis and references data are necessary to guarantee that the radiometric value encountered corresponds that associated land cover. This study can contribute in the future with many applications such as 3D reconstruction, image matching, object recognition, etc.
%9 Processamento de imagens
%@language pt
%3 p0729.pdf