基于LBP与SVM的马铃薯芽眼识别  

Potato Bud Recognition Based on LBP and SVM

作　　者：张金敏[1] 杨添玺 ZHANG Jin-min;YANG Tian-xi(Lanzhou Jiaotong University,Lanzhou 730070,China)

机构地区：[1]兰州交通大学,甘肃兰州730070

出　　处：《山东农业大学学报（自然科学版）》2020年第4期744-748,共5页Journal of Shandong Agricultural University：Natural Science Edition

摘　　要：芽眼是马铃薯种植的关键,芽眼识别是种薯智能切块的先决条件。当前,马铃薯种植已实现一定程度的机械化,但在种植中,种薯切块仍由人工完成。本文为提高切块效率,对马铃薯芽眼识别进行研究,通过摄像头采集马铃薯图像,首先进行中值滤波,接着采用大津法进行图像分割;然后采用局部二值模式分别对新旧种薯芽眼和非芽眼区域进行特征提取,最后采用支持向量机进行样本特征训练,通过对不同样本进行实验,综合识别率达97.33%。综上实验,基于局部二值模式进行特征提取与支持向量机进行分类的方法,种薯芽眼识别率达到了令人满意的水平,为种薯智能切块奠定良好基础。Bud eye is the key to potato planting,and bud eye recognition is a prerequisite for intelligent cutting of seed potato.At present,potato planting has achieved a certain degree of mechanization,but in the planting,potato cutting is still done manually.In order to improve the efficiency of slicing,the recognition of potato bud eyes was studied in this paper,the image of potato was collected by camera,first median filter was carried out,followed by image segmentation with the Otsu method,then the local binary model was used to extract the new and old potato bud and non-bud eye regions respectively,finally,the support vector machine was used for sample feature training.Through experiments on different samples,the comprehensive recognition rate can reach 97.33%.In conclusion,the method of feature extraction and classification by support vector machine based on local binary model has reached a satisfactory level,which lays a good foundation for intelligent potato cutting.

关 键 词：马铃薯 芽眼识别 局部二值模式 支持向量机 

分 类 号：S532[农业科学—作物学]