农机化研究
農機化研究
농궤화연구
JOURNAL OF AGRICULTURAL MECHANIZATION RESEARCH
2013年
4期
140-143
,共4页
高光谱成像技术%灰度共生矩阵%支持向量机%鉴别%碧螺春茶
高光譜成像技術%灰度共生矩陣%支持嚮量機%鑒彆%碧螺春茶
고광보성상기술%회도공생구진%지지향량궤%감별%벽라춘다
hyperspectral imaging technique%gray level co-occurrence matrix%support vector machine%discrimination%Biluochun tea
提出了基于漫反射式高光谱成像技术结合模式识别的碧螺春茶真伪鉴别方法。试验以漫反射式高光谱成像系统采集碧螺春茶样图像数据；通过主成分分析优选特征图像，并利用灰度共生矩阵从中各提取4个纹理特征参量；最后，利用支持向量机的模式识别方法构建茶叶真伪鉴别模型。训练和预测集的判别率分别达到100％和96．25％。研究表明，利用漫反射式高光谱成像技术结合支持向量机模式识别方法鉴别碧螺春茶品质真伪是可行的。
提齣瞭基于漫反射式高光譜成像技術結閤模式識彆的碧螺春茶真偽鑒彆方法。試驗以漫反射式高光譜成像繫統採集碧螺春茶樣圖像數據；通過主成分分析優選特徵圖像，併利用灰度共生矩陣從中各提取4箇紋理特徵參量；最後，利用支持嚮量機的模式識彆方法構建茶葉真偽鑒彆模型。訓練和預測集的判彆率分彆達到100％和96．25％。研究錶明，利用漫反射式高光譜成像技術結閤支持嚮量機模式識彆方法鑒彆碧螺春茶品質真偽是可行的。
제출료기우만반사식고광보성상기술결합모식식별적벽라춘다진위감별방법。시험이만반사식고광보성상계통채집벽라춘다양도상수거；통과주성분분석우선특정도상，병이용회도공생구진종중각제취4개문리특정삼량；최후，이용지지향량궤적모식식별방법구건다협진위감별모형。훈련화예측집적판별솔분별체도100％화96．25％。연구표명，이용만반사식고광보성상기술결합지지향량궤모식식별방법감별벽라춘다품질진위시가행적。
A new idea that the famous brand tea ’s authenticity was discriminated by hyperspectral imaging technique was proposed in this work.Hyerpspectral imaging system based on diffuse reflection was developed for Biluochun tea ’s data acquisition.Feature wavebands images were selected from hyperspectral data by principal component analysis and four characteristic variables were extracted from feature wavebands images by texture analysis based on the gray -level co-oc-currence matrix.Finally, Discrimination model was developed by support vector machine (SVM).The back-discrimina-tion rate of model is 100%in the calibration set, and the discrimination rate of model is 96.25%in the prediction set. This study shows that it is feasible discriminate Biluochun tea ’s authenticity by hyperspectral imaging technique coupled with SVM pattern recognition.
