An Automated Galaxy Recognition and Redshift Measurement System for Low-resolution Spectra

For the vast amounts of spectra produced by LAMOST, the pipeline based on the PCAZ method with whole spectral features failed for galaxy spectra recognition. There are three reasons for this: bad flux calibration of most galaxy spectra, low SNR, and the key algorithm of the pipeline based on the overall spectral features. In this paper, we present a new automated method for galaxy spectra recognition and redshift measurement. The method, named GM, includes three units: 1) Extracting the spectral lines through continuum normalization; 2) Galaxy spectra template construction based on k-mean clustering; 3) Template matching for galaxy spectra classification and redshift measurement. The experiment verified this is feasible for the LAMOST galaxy spectra: a correct rate of > 80% for the data with SNR_g > 4, SNR_r > 6. Compared with the redshifts from SDSS, the systematic error of our method is 0, and the standard deviation of the error is 0.0002.