计算机应用 ›› 2018, Vol. 38 ›› Issue (4): 960-964.DOI: 10.11772/j.issn.1001-9081.2017092342

• 人工智能 • 上一篇    下一篇

基于AdaBoost.M2和神经模糊系统的植物识别算法

雷建椿, 何金国   

  1. 中央民族大学 理学院, 北京 100081
  • 收稿日期:2017-09-28 修回日期:2017-11-23 出版日期:2018-04-10 发布日期:2018-04-09
  • 通讯作者: 雷建椿
  • 作者简介:雷建椿(1990-),男,福建福安人,硕士研究生,主要研究方向:图像与信息处理、数据挖掘;何金国(1973-),男,江西上饶人,副教授,博士,主要研究方向:图像处理、区间集、智能信息处理、数据挖掘。

Plant recognition algorithm based on AdaBoost.M2 and neural fuzzy system

LEI Jianchun, HE Jinguo   

  1. College of Science, Minzu University of China, Beijing 100081, China
  • Received:2017-09-28 Revised:2017-11-23 Online:2018-04-10 Published:2018-04-09

摘要: 为提高传统神经模糊系统(NFS)在植物识别领域对于相似植物样本的识别能力,提出了AdaBoost.M2-NFS算法。该算法首先对传统NFS进行改进以便融合,然后将新NFS与AdaBoost.M2结合得到AdaBoost.M2-NFS新模型。在Iris数据集上实验结果表明:新模型与单个NFS相比,识别率增加了3.33个百分点;与线性支持向量机(SVM)相比,识别率增加了1.11个百分点;与Softmax相比,识别率增加了3.33个百分点。根据敏感性和特异性分析可知,所提模型对于线性不可分数据分类效果比对线性可分数据分类效果好;同时,由于AdaBoost.M2的改进,使得所提算法在植物识别领域具备快速成型和高泛化能力。

关键词: AdaBoost.M2, 神经模糊系统, 植物识别, 支持向量机

Abstract: An AdaBoost.M2-NFS model was presented to improve the recognition rate of traditional Neural Fuzzy System (NFS) towards similar plants. The traditional NFS was improved for fusion, and then the new NFS was combined with AdaBoost.M2 to get a new AdaBoost.M2-NFS model. Experimental results show that the new model increases the recognition rate by 3.33 percentage points compared with the single NFS; compared with the linear Support Vector Machine (SVM), its recognition rate increases by 1.11 percentage points; compared with Softmax, its recognition rate increases by 3.33 percentage points. Based on sensitivity and specificity analysis, the non-linear data can get better classification result than the linear data by the proposed algorithm. At the same time, due to the improvement of AdaBoost.M2, the new algorithm has the advantages of modeling quickly and high generalization ability in the field of plant recognition.

Key words: AdaBoost.M2, Neural Fuzzy System (NFS), plant recognition, Support Vector Machine (SVM)

中图分类号: