RESEARCH ON THE PARAMETERS OF MULTISPECTRAL IMAGES AND THE EFFECTIVENESS OF CLASSIFYING NATURAL AND ANTHROPOGENIC OBJECTS
V.A. Aksenov, S.I. Orlov, O.I. Potaturkin, S.B. Uzilov
Institute of Automation and Electrometry, Siberian Branch, Russian Academy of Sciences, Novosibirsk, Russia
Keywords: classification accuracy, natural and anthropogenic objects, spectral intervals, multispectral features, feature systems, processing of multispectral images
Abstract
The systems of three features for the classification of natural and anthropogenic objects in the observation field using spectral channels of various widths from 5 to 90 nm have been studied. In comparison with previously obtained results, the effectiveness of using feature systems formed in the most informative intervals of the visible and near-IR ranges has been experimentally demonstrated. Systems of three features based on the formation of multispectral images in wide intervals of the visible and narrow intervals of the near-IR ranges have been identified, which ensure high accuracy in classifying objects in real scenes.
|
