摘要
为保障小花清风藤的药材质量,建立基于HOG特征与BP神经网络相结合的小花清风藤叶片智能鉴别模型。采用方向梯度直方图(HOG)特征提取算子提取小花清风藤与簇花清风藤叶片的HOG特征,结合BP神经网络理论构建基于HOG特征的小花清风藤叶片智能鉴别神经网络模型。结果表明:该模型对小花清风藤叶片的平均识别率为97.60%,簇花清风藤为98.61%,对任意样本为100%。建立的基于HOG特征与BP神经网络相结合的小花清风藤叶片智能鉴别模型适合于小花清风藤叶片鉴别。
In order to ensure the quality of the medicinal materials of Sabia parviflora,an intelligent identification model of S.parviflora was established based on integrated HOG characteristics and BP neural network.The HOG features of the leaves of S.parvifloraand Sabia swinhoei were extracted by using the direction gradient histogram(HOG)feature extraction operator.Integrated with BP neural network theory,the intelligent neural network model of S.parviflora was constructed based on HOG features.Results:The average recognition rate of the model was 97.60%for the leaves of S.parviflora,98.61%for that of S.swinhoei,and 100%for any sample.The established intelligent identification model of S.parviflora based on the combination of HOG features and BP neural network is suitable for the identification of S.parvifloraleaves.
引文
[1]中国科学院中国植物志编辑委员会.中国植物志:第47卷(第1分册)[M].北京:科学出版社,1985:73-75.
[2]高秀丽.Flora of China:第12卷[M].北京:科技出版社,2007:25-42.
[3]罗迎春,孙庆文.贵州民族常用天然药物(第2册)[M].贵阳:贵州科技出版社,2013:174.
[4]温迪,孙庆文,潘国吉,等.清风藤属药用植物研究进展[J].贵州科学,2016,34(3):25-31.
[5]孙庆文,何顺志.贵州大戟科、鼠李科等4个科药用植物资源的研究[J].贵州科学,2005,23(4):54-58.
[6]孙庆文.贵州民族常用天然药物[M].贵阳:贵州科技出版社,2013.
[7]潘国吉,徐文芬,孙庆文,等.小花清风藤总黄酮的提取工艺及其含量的测定[J].华西药学杂志,2018(3):295-300.
[8]潘国吉,孙庆文,徐文芬,等.小花清风藤中总皂苷含有量测定[J].中成药,2018,40(6):1418-1421.
[9]温迪,孙庆文,陈敬忠,等.清风藤属药用植物石细胞特征及其鉴定学意义[J].中药材,2016,39(9):1989-1992.
[10]孙庆文,齐维娜,王波.清风藤属药用植物叶表皮特征及其鉴定学意义[J].中药材,2013,36(10):1594-1598.
[11]王波,齐维娜,孙庆文.4种清风藤属药用植物的外观形态与显微特征[J].贵州农业科学,2014,42(9):75-80.
[12]郭文凯,孙庆文,周晓琴.小花清风藤扦插繁殖技术研究[J].贵州农业科学,2018,46(12):109-112.
[13]TA爦KIRAN M,AM Z G.Offline signature identification via HOG features and artificial neural networks[C]//IEEE.International Symposium on Applied Machine Intelligence and Informatics.IEEE,2017:83-86.
[14]何正风.MATLAB R2015b神经网络技术[M].北京:清华大学出版社,2016.
[15]曹永峰,赵燕君.基于GA-BP神经网络的计算机智能化图像识别技术探究[J].应用激光,2017(1):139-143.
[16]周舒冬,李倚岳,李丽霞,等.基于LM-BP神经网络的穿心莲药材分类识别[J].时珍国医国药,2009,20(3):553-555.