Mineral identification of hyperspectral images based on the stability of the spectral characteristic parameters

Yongfei Che; Yingjun Zhao

doi:10.1117/12.2268382

Translator Disclaimer

8 March 2017 Mineral identification of hyperspectral images based on the stability of the spectral characteristic parameters

Yongfei Che, Yingjun Zhao

Author Affiliations +

Proceedings Volume 10255, Selected Papers of the Chinese Society for Optical Engineering Conferences held October and November 2016; 1025528 (2017) https://doi.org/10.1117/12.2268382
Event: Selected Papers of the Chinese Society for Optical Engineering Conferences held October and November 2016, 2016, Jinhua, Suzhou, Chengdu, Xi'an, Wuxi, China

Abstract

Spectral stability characteristic parameter analysis is the basis of all the quantitative information extraction in hyperspectral image. The results show that the stability of the spectral parameters used in the spectral identification has a great influence on the efficiency of mineral identification. A mineral recognition method for hyperspectral remote sensing image based on spectral stability characteristic parameter is introduced. First, reference spectrum spectral peaks and valleys positions were extracted, then calculates the measured spectra corresponding spectral wavelength and reference spectrum of each with a characteristic peak and valley of the correlation coefficient, basis of comparison of two spectral similarity to determine the matching effect of the two spectra, in order to achieve the best mineral identification precision and accuracy. Gansu Beishan Shijinpo gold mining area as an example, the mineral identification map was obtained. After field verification, it was confirmed that the method has higher accuracyon the mineral recognition.

Citation Download Citation

Yongfei Che and Yingjun Zhao "Mineral identification of hyperspectral images based on the stability of the spectral characteristic parameters", Proc. SPIE 10255, Selected Papers of the Chinese Society for Optical Engineering Conferences held October and November 2016, 1025528 (8 March 2017); https://doi.org/10.1117/12.2268382

ACCESS THE FULL ARTICLE