Dong-mei Qin | PubMed Health

PubMed2004

[Automated classification of celestial spectra based on support vector machines].

The main objective of an automatic recognition system of celestial objects via their spectra is to classify celestial spectra and estimate physical parameters automatically. This paper proposes a new automatic classification method based on support vector machines to separate non-active objects from active objects via their spectra. With low SNR and unknown red-shift value, it is difficult to extract true spectral lines, and as a result, active objects can not be determined by finding strong spectral lines and the spectral classification between non-active and active objects becomes difficult. The proposed method in this paper combines the principal component analysis with support vector machines, and can automatically recognize the spectra of active objects with unknown red-shift values from non-active objects. It finds its applicability in the automatic processing of voluminous observed data from large sky surveys in astronomy.

Algorithms↗

PubMed2003

[A PCA based efficient stellar spectra classification method].

Stellar spectra classification is an indispensable part of any workable automated recognition system of celestial bodies. This paper introduces an efficient method of automated classification of stellar spectra based on the principal component analysis (PCA). The method consists of two parts. In the first part, the eigen-matrix is built by a standard PCA technique where only the first two eigenvectors are selected due to their predominance. More specifically, the first two eigenvalues are found to always represent more than 95% of the total sum of all the eigenvalues, and much larger than others in our all experiments. The principal component space of stellar (V1, V2) then is constructed from the first two eigenvectors. In the second part, namely classification part, an unknown spectrum X is first mapped to a 2D space with the two coordinates defined respectively as (V1T X, V2T X), then the nearest neighbor approach in this 2D space is employed to determine the spectral type as well as luminosity class of the input spectrum. The experimental results show that our new method can achieve comparable performance with that by the standard MK spectral types classification criterion, which is regarded as a benchmark in astronomy field. Thanks to its high efficiency, our new method appears promising especially for the processing of spectra in large quantities collected from large survey projects, such as LAMOST project in our country.

English Abstract↗