CAJ | 학술논문

利用可见／近红外光谱技术对梨枣轻微损伤的分类判别建模方法进行研究。分别采用 PLS－LDA （线性）和LS－SVM （非线性）建立判别模型，分析比较不同预处理方式和建模波段对模型精度的影响。结果表明：经9点平滑预处理后的短波近红外（780～1100nm）PLS－LDA模型判别效果最佳，校正集和预测集的正确识别率分别达到97．8％和96．7％。
이용가견／근홍외광보기술대리조경미손상적분류판별건모방법진행연구。분별채용 PLS－LDA （선성）화LS－SVM （비선성）건립판별모형，분석비교불동예처리방식화건모파단대모형정도적영향。결과표명：경9점평활예처리후적단파근홍외（780～1100nm）PLS－LDA모형판별효과최가，교정집화예측집적정학식별솔분별체도97．8％화96．7％。
Classification discrimination models of subtle bruise “Lizao” jujubes were studied by using visible/near-infra-red spectroscopy .The least squares-support vector machine ( LS-SVM) and partial least squares-line discriminant anal-ysis( PLS-LDA) were applied to develop classification models , which based on different spectra pretreatments and differ-ent modeling wavelength bands .The results indicated that the best discrimination model was PLS -LDA model using Savitzky-Golay of 9 in the range of 780~1100nm, for the calibration and prediction set of PLS-LDA model, the dis-crimination rate was 97.8% and 96.7%, respectively.