CAJ | 학술논문

含有隐蔽性害虫卵和幼虫的小麦籽粒，其生物光子辐射信号与正常小麦有显著不同。为此，针对大量含虫和不含虫小麦籽粒的生物光子辐射自发发光信号，选择均值、标准差以及光子统计熵等作为特征参数，构造分类器对受试小麦样品进行分类。实验结果表明，与最近邻分类算法、KNN 分类算法相比，加权 KNN 分类器具有良好的分类效果，正确率达到92．5％，研究成果为粮食作物隐蔽性虫害的预报和检测提供了一种新的思路。
함유은폐성해충란화유충적소맥자립，기생물광자복사신호여정상소맥유현저불동。위차，침대대량함충화불함충소맥자립적생물광자복사자발발광신호，선택균치、표준차이급광자통계적등작위특정삼수，구조분류기대수시소맥양품진행분류。실험결과표명，여최근린분류산법、KNN 분류산법상비，가권 KNN 분류기구유량호적분류효과，정학솔체도92．5％，연구성과위양식작물은폐성충해적예보화검측제공료일충신적사로。
The ultra weak bioluminescence signals of wheat kernels with the eggs and larvae of hidden insect are signifi -cantly different from that of normal wheat kernels .In this paper , the ultra weak bioluminescence signals of wheat with hidden insect and normal wheat are measured firstly .Then such values as mean , standard deviation and photon statistical entropy are chosen as the feature parameters .And some classifiers to distinguish the different states of wheat kernels are constructed .Through comparing the experiment results of using the designed classifiers with nearest neighbor classification algorithm , KNN classification algorithm and weighted KNN classification algorithm , it shows that the weighted KNN clas-sifier classification effect is the best , and the classification accuracy is up to 92 .5%.This paper provides a research way to detect whether food crops has hidden insects in the sense of biophoton analytical technology .