INVESTIGATION OF THE HYPERSPECTRAL DATA CLASSIFICATION EFFICIENCY UNDER CONSTRAINTS ON THE BIT DEPTH OF QUANTIZATION, NUMBER OF SPECTRAL CHANNELS, AND SPATIAL RESOLUTION
V.V. Shipko1, S.M. Borzov2
1Military Educational Scientific Center of the Air Force "Air Force Academy named after Prof. N.E. Zhukovsky and Yu.A. Gagarin,", Voronezh, Russia 2Institute of Automation and Electrometry, Siberian Branch, Russian Academy of Sciences, Novosibirsk, Russia
Keywords: hyperspectral images, classification, quantization, spatial resolution, spectral channels
Abstract
The possibility of creating a fast-response framing hyperspectrometer functioning as a conventional camera is studied. It is considered how much the classification accuracy will have to be sacrificed for the implementation of the idea of creating a compact and easy-to-use hyperspectral camera with rapidly tunable parameters. Experimental studies of the classification accuracy depending on the number and method of selection of spectral channels, spatial resolution, and bit depth of quantization of the signal brightness are carried out on a test hyperspectral image. It is shown that, though the resulting accuracy decreases, it still remains acceptable for a wide range of practical tasks.
|