欢迎访问《中国农学通报》，
中国农学通报

中国农学通报 ›› 2025, Vol. 41 ›› Issue (4): 156-164.doi: 10.11924/j.issn.1000-6850.casb2024-0295

• 农业信息·科技教育 • 上一篇    

基于机器视觉的菜豆种子品种智能识别与分类

李淑佳1(), 孙来军1(), 蒙亚浩1, 王祎辰1, 李晓旭2, 冯国军3, 杨凤艳4   

  1. 1 黑龙江大学电子工程学院，哈尔滨 150080
    2 中国移动通信集团山东有限公司淄博分公司，山东淄博 255020
    3 黑龙江大学现代农业与生态环境学院，哈尔滨 150080
    4 黑龙江农业工程职业学院，哈尔滨 150080
  • 收稿日期:2024-05-07 修回日期:2024-11-13 出版日期:2025-01-23 发布日期:2025-01-23
  • 通讯作者:
    孙来军，男，1977年出生，山东嘉祥人，教授，博士，研究方向：无损检测技术、农业检测、设备状态检测与故障诊断。通信地址：150080 黑龙江省哈尔滨市南岗区学府路74号 黑龙江大学A8栋504室，E-mail：。
  • 作者简介:

    李淑佳，女，1999年出生，山东东营人，硕士，研究方向为智能检测与模式识别。通信地址：150080 黑龙江省哈尔滨市南岗区学府路74号 黑龙江大学电子工程学院，E-mail：。

  • 基金资助:
    黑龙江省自然科学基金项目“基于多源信息融合与SVM的水稻品质分类与年份迭代优化研究”(SS2021C005); 黑龙江省重点研发计划项目“高端电子系统板级和设备级的多余物高精密检测技术研发”(2022ZX03A06)

Intelligent Identification and Classification of Common Bean Seed Varieties Based on Machine Vision

LI Shujia1(), SUN Laijun1(), MENG Yahao1, WANG Yichen1, LI Xiaoxu2, FENG Guojun3, YANG Fengyan4   

  1. 1 College of Electronic Engineering, Heilongjiang University, Harbin 150080
    2 Zibo Branch, China Mobile Communications Group Shandong Co., Ltd., Zibo, Shandong 255020
    3 College of Modern Agriculture and Ecological Environment, Heilongjiang University, Harbin 150080
    4 Heilongjiang Agricultural Engineering Vocational College, Harbin 150080
  • Received:2024-05-07 Revised:2024-11-13 Published:2025-01-23 Online:2025-01-23

RichHTML

1

PDF (PC)

16

摘要：

为基于机器视觉(MV)设计一种低成本、高效且无损的菜豆种子识别、分类的方法，采集6个品种2751粒菜豆种子的图像信息，在对图像进行二值化、颜色提取、形态学操作等图像处理的基础上，提取包括颜色特征、纹理特征以及几何特征在内的9种特征作为分类的依据，分别建立K近邻(KNN)、随机森林(RF)、支持向量机(SVM)3种分类模型，对菜豆种子的品种进行分类。对比3种分类模型的混淆矩阵、准确率及F1值后发现，SVM模型的分类效果上最优，其分类准确率和F1值分别达到97.7%、0.977。研究表明，利用MV可以实现对菜豆种子的精准识别和分类。

关键词: 菜豆, 分类, 机器视觉, 图像处理, 智能识别, 机器学习, 支持向量机, 特征提取

Abstract:

The aim of this study was to design a low-cost, efficient and non-destructive method for identifying and classifying common bean seeds based on machine vision (MV). In this study, image information of 2751 seeds of six varieties of common beans was collected, and based on image processing such as binarization, color extraction and morphological operations, nine features including color features, texture features and geometric features were extracted as the basis of classification, and K-nearest neighbor (KNN), random forest (RF), and support vector machine (SVM) classification models were established to classify the varieties of bean seeds. After comparing the confusion matrix, accuracy and F1 value of the three classification models, it was found that the SVM model outperformed the other two classification models, with a classification accuracy and F1 value of 97.7% and 0.977, respectively. The results of the study show that accurate identification and classification of common bean seeds can be achieved using MV.

Key words: common bean, classification, machine vision, image processing, intelligent recognition, machine learning, support vector machine, feature extraction