Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa, 2019, Vol. 16, No. 6, pp. 201-208
Classification of water surfaces by spectra measured from the ISS during the Uragan space experiment
M.Yu. Belyaev
1 , B.I. Belyaev
2 , L.V. Katkovsky
2 , A.O. Martinov
2 , E.E. Sarmin
1 , V.A. Siliuk
2 , A.V. Chumakov
2 1 S.P. Korolev Rocket and Space Public Сorporation “Energia”, Korolev, Russia
2 A.N. Sevchenko Institute of Applied Physical Problems of Belorusian State University, Minsk, Republic of Belarus
Accepted: 01.10.2019
DOI: 10.21046/2070-7401-2019-16-6-201-208
The paper describes the results of processing of multi-temporal measurements (images and spectra) of the Earth surface made by the scientific equipment Photospectral System (PhSS) from the International Space Station (ISS) during the Uragan space experiment. A new classification methodology for spectral data based on the principal component method and correlation analysis has been developed. The methodology is effective for the classification of objects whose reflection spectra are close to each other both in form and in absolute values. In contrast to the traditional application of principal component transformation, where the first few principal components are used for analysis, in the proposed method the first component that has the greatest variance is excluded from the analysis. The inverse transformation into the spectral space is performed, which allows taking into account minor differences in the spectra of objects. In addition, a visualization of the classified spectra in three-dimensional space based on the correlations of the entire sample of spectra is proposed. Using the proposed methodology, classifications of the water surfaces (seas and oceans) are done on the basis of PhSS data. The results of classification demonstrate the possibility of dividing water surfaces according to their reflection spectra, which indicates the existence of characteristic spectral features of the waters of different seas.
Keywords: remote sensing, spectra, correlation method, principal component analysis, classification
Full textReferences:
- Belyaev B. I., Belyaev M. Yu., Desinov L. V., Katkovskii L. V., Krot Yu. A., Sarmin E. E. (2014a), Rezul’taty ispytanii fotospektral’noi sistemy na MKS (Photospectral system test results on ISS), Issledovanie Zemli iz kosmosa, 2014, No. 6, pp. 27–39.
- Belyaev B. I., Belyaev M. Yu., Desinov L.V, Katkovskii L. V., Sarmin E. E. (2014b), Obrabotka spektrov i izobrazhenii s fotospektral’noi sistemy v kosmicheskom eksperimente “Uragan” na MKS (Spectra and images processing from Photospectral system in space experiment “Hurricane” on the ISS), Issledovanie Zemli iz kosmosa, 2014, No. 6, pp. 54–65.
- Lagutin B. M., Naglyadnaya matematicheskaya statistika (Visible mathematical statistics), Moscow: BINOM. Laboratoriya znanii, 2007, 472 p.