ISSN 2070-7401 (Print), ISSN 2411-0280 (Online)
Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa


Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa, 2013, Vol. 10, No. 3, pp. 75-84

Two-step algorithm for the classification of hyperspectral data in the space of the spectral brightness coefficients of aerial photography

V.N. Ostrikov , S.I. Smirnov , V.V. Mikhailov 
LUCH Construction Bureau, St.-Petersburg Division
In the paper the problem of the classification of hyperspectral data taken in the space of coefficients of the spectral brightness is observed. It is assumed that each object in the hypercube belongs to a spectral class contained in the database, formed by ground-based measurements of the spectral characteristics of objects. The inputs to the algorithm are the hyperspectral data, passed calibration, filtering of regular and random noise, geometric correction, and a database of measured spectral brightness. Classification is carried out in two steps. At the first step the classes that obviously can not belong to the object are screened, using a rather “rough” measure of proximity. The procedure can significantly reduce the amount of data to be processed, which will positively affect the speed of the algorithm. In the second step from a set of classes, “remaining” after the pre-classification, the closest to the object elements of the database are selected using a more sensitive to the spectral difference metric (e.g., Terebizh’s metric, Euclid metric). The algorithm is tested on hyperspectral survey data obtained from aircraft carrier under different conditions of observation. The robustness of the algorithm was revealed in a wide range of signal-to-noise ratio. A comparison of the classification results on real images was held.
Keywords: классификация гиперспектральных данных, метрика Теребижа, кластеризация, classification of hyperspectral data, the Terebizh’s metric, clustering
Full text


  1. Zagoruiko N.G., Prikladnye metody analiza dannykh i znanii (Applied methods of data and knowledge analysis), Novosibirsk: IM SO RAN, 1999, 270 p.
  2. Ostrikov V.N., Plakhotnikov O.V., Kirienko A.V., Desyataya vserossiiskaya otkrytaya konferentsiya “Sovremennye problemy distantsionnogo zondirovaniya Zemli iz kosmosa” (Current Problems in Remote Sensing of the Earth from Space), Proc. 10th All-Russia Open Conference, 2012.
  3. Terebizh V.Yu., Vvedenie v statisticheskuyu teoriyu obratnykh zadach (Introduction to statistical theory of inverse problems), Moscow: Fizmatlit, 2005, 376 p.
  4. Shovengerdt R.A., Remote Sensing. Models and Methods for Image Processing, Amsterdam, 2007, 515 p.
  5. Lloyd P., Least square quantization in PCM’s, Bell Telephone Laboratories Paper, 1957.
  6. MacQueen J., Some methods for classification and analysis of multivariate observations, Proc. 5th Berkeley Symp. on Math. Statistics and Probability, 1967, pp. 281–297.
  7. Steinhaus H., Sur la division des corps matériels en parties, Bulletin of the Polish Academy of Sciences, C1. III, Vol. IV, 1956, pp. 801–804.