COMPLEXING OF CLASSIFIERS IN THE PROBLEM OF THEMATIC PROCESSING OF HYPERSPECTRAL AEROSPACE IMAGES
E. V. Dmitriev1, V. V. Kozoderov2, A. O. Dement'ev1, A. N. Safonova3
1Institute of Computational Mathematics, Russian Academy of Sciences, 119333, Moscow, ul. Gubkina, 8
2Moscow State University, 119991, Moscow, Leninskie Gory, 1
3Institute of Space and Information Technologies, 660074, Krasnoyarsk, ul. Academika Kirenskogo, 26, korp. 1
Keywords: дистанционное зондирование, гиперспектральные изображения, распознавание образов, тематическая обработка, классификация, remote sensing, hyperspectral images, pattern recognition, thematic processing, classification
Abstract
This paper describes the problem of thematic processing of hyperspectral aerospace images. The proposed technique is based on a decoding algorithm, allows constructing an effective algorithm of multiclass classification based on a combination of standard algorithms of binary classification of various complexity. The choice of the configuration of the classification algorithm is substantiated. Test calculations based on model and real data are used to show the effectiveness of the proposed approach to recognition of objects by hyperspectral images.
|
