Neighboring Discriminant Component Analysis for Asteroid Spectrum Classification

With the rapid development of aeronautic and deep space exploration technologies, a large number high-resolution asteroid spectral data have been gathered, which can provide diagnostic information for identifying different categories asteroids as well their surface composition mineralogical properties. However, owing to noise observation systems ever-changing external environments, observed always contain outliers exhibiting indivisible pattern characteristics, will bring great challenges precise classification asteroids. In order alleviate problem improve separability accuracy kinds asteroids, this paper presents novel Neighboring Discriminant Component Analysis (NDCA) model spectrum feature learning. The key motivation is transform from into subspace wherein negative effects be minimized while category-related valuable knowledge in explored. effectiveness proposed NDCA verified on real-world reflectance spectra measured over wavelength range 0.45 2.45 μm, promising performance has achieved by combination with classifier models, such nearest neighbor (NN), support vector machine (SVM) extreme learning (ELM).