基于双曲切比雪夫逼近的光谱模板构造

A Stellar Spectrum Template Construction Method Based on Two Curve Chebyshev Approximation

我国正在实施的大型巡天项目(LAMOST),迫切需要一套恒星光谱自动识别与分类系统。恒星光谱的自动识别与分类很大程度上依赖于恒星光谱模板库的质量。恒星光谱分类模板库中模板光谱数量越少,恒星光谱的自动处理速度越快。论文给出了一种基于切比雪夫双曲线逼近(two curve Chebyshev approximation)分段式构造恒星光谱模版的方法。该方法由以下主要步骤组成:1)将光谱数据分段,分为多个光谱段;2)基于切比雪夫双曲线逼近提取每类每段光谱数据谱线的形状特征;3)由切比雪夫逼近得到的每类光谱的形状特征与对应的该类光谱的标准形状特征进行比较。实验结果表明该方法构造的恒星光谱模版库提高了现有恒星自动识别与分类系统的精度与可信度。

The large scale sky survey project(LAMOST)being implemented in our country urgently needs a set of automatic stellar spectrum identification and classification system.The automatic identification and classification of stellar spectra largely de⁃pends on the quality of the stellar spectral template library.The fewer template spectra in the stellar spectrum classification template library,the faster the automatic processing speed of stellar spectra.This paper presents a method of constructing a stellar spectrum template in segments based on the two curve Chebyshev approximation.The method consists of the following main steps.The first is to Segment the spectral data into multiple spectral segments.The second is to extract the shape features of each type of spectral data line based on two curve Chebyshev approximation.The third is to compare the shape feature of each type of spectrum obtained by Chebyshev approximation with the corresponding standard shape feature of that type of spectrum.The experimental results show that the stellar spectrum template library constructed by this method improves the accuracy and credibility of the existing automatic star identification and classification system.