高光谱成像的油菜和杂草分类方法

摘要

A classification method of oilseed rape and weeds based on hyperspectral information was put forward.Standard nor-mal variate transformation (SNV),de-trending,multiplicative scatter correction (MSC),moving average (MA),savitzky-golay smoothing(SG),baseline and normalize were applied to data preprocess.Principal component analysis loadings (PCA loadings), x-loading weights,regression coefficient (RC)and successive projection algorithm (SPA)were used to extract feature wave-lengths.Partial least-squares discriminant analysis (PLS-DA),extreme learning machine (ELM)and support vector machine (SVM)were employed to establish classification models.The overall results shows that the ELM models with the selected wave-lengths of PCA loadings,x-loading weights and SPA based on de-trending preprocessed spectra has obtained the best results, with 100％ classification accuracy for both the calibration set and the prediction set.The index of average classification accuracy is introduced to evaluate classification models accuracy under different experimental time.The results indicates that it is feasible to use near-infrared hyperspectral imaging to identify the oilseed rape and weeds.%利用高光谱成像技术结合化学计量学方法对油菜中的杂草进行分类识别.采用近红外高光谱技术,通过正态变量变换(SNV)、去趋势化(De-trending)、多元散射校正(MSC)、移动平均平滑法(MA)、多项式卷积平滑法(SG)、基线校正(baseline)及归一化(normalize)算法对光谱数据进行预处理,采用主成分载荷(PCA loadings)、载荷系数法(x-LW)、回归系数法(RC)、连续投影算法(SPA)分别进行特征波长提取,采用偏最小二乘判别分析(PLS-DA)、极限学习机(ELM)和支持向量机(SVM)建立分类模型.结果表明,基于De-trending预处理,通过PCA loadings,x-loading weights及SPA特征波长提取方法,基于极限学习机ELM算法建立的模型取得了最优的分类效果,建模集和预测集的分类精度均达到100％,另引入平均分类精度的指标,发现不同试验时间下,模型分类精度变化不大,表明应用近红外高光谱成像技术对油菜和杂草进行分类是可行的.