基于LDASVM的小麦质地检测方法研究  

Research on Wheat Texture Detection Method Based on LDASVM

作　　者：赵薇 赵雪妮[1] 康凯 刘长斌 罗斌 张晗 Zhao Wei;Zhao Xueni;Kang Kai;Liu Changbin;Luo Bin;Zhang Han(College of Mechanical&Electrical Engineering,Shaanxi University of Science&Technology,Xi’an 710021;Intelligent Equipment Research Center,Beijing Academy of Agriculture and Forestry Sciences,Beijing 100097;Information Technology Research Center,Beijing Academy of Agriculture and Forestry Sciences,Beijing 100097)

机构地区：[1]陕西科技大学机电工程学院,西安710021 [2]北京市农林科学院智能装备技术研究中心,北京100097 [3]北京市农林科学院信息技术研究中心,北京100097

出　　处：《中国粮油学报》2023年第1期146-152,共7页Journal of the Chinese Cereals and Oils Association

基　　金：国家重点研发计划项目(2017YFD0701205);北京市农林科学院青年基金项目(QNJJ202104)。

摘　　要：研究基于透射光图像的小麦质地检测方法,使用工业相机采集14种小麦种子的透射光图像,通过图像处理技术获取整粒小麦、胚乳和种胚代表性区域,并提取对应区域的颜色特征数据。分别运用PCA和LDA进行数据降维,并将降维前后的数据与支持向量机(SVM)、K近邻算法(KNN)和决策树模型(DT)3种分类器相结合建立分类模型,对不同品种小麦质地进行分类识别研究。结果表明:利用图像处理技术提取透射光全部特征,建立的LDA_SVM模型分类正确率可以达到97%以上,证明透射光图像下通过机器学习对不同质地小麦快速分类鉴别是可行的。In the present study, a method of wheat texture detection based on transmitted light images was investigated. Transmitted light images of 14 wheat seeds were collected by industrial camera. Representative regions of whole wheat, endosperm and seed embryo were obtained by image processing techniques, and color feature data of corresponding regions were extracted. Then, PCA and LDA were applied to dimensionality reduction of the data respectively, and the data before and after dimensionality reduction were combined with support vector machine(SVM), k-nearest neighbor algorithm(KNN) and decision tree model(DT) to establish classification models to study the classification and recognition of different varieties of wheat textures. The results showed that the LDA_SVM model established by extracting all the features of transmitted light using image processing techniques could reach over 97% correct classification rate. In this regard, it was feasible to quickly classify and identify wheat with different texture by machine learning under transmitted light images.

关 键 词：透射光 小麦质地 角质率 机器视觉 机器学习 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] S512.1[自动化与计算机技术—计算机科学与技术]