摘要
因洱海湿地昆虫具有形态微小、不易识别的特点,为提高昆虫数据智能化处理和分类效率,设计了一种微小昆虫智能识别方法。该方法通过SVM-AdaBoost机器学习模型实现了昆虫图像的分类学习,并基于IOS移动平台,采用MVC(Model View Controller)设计模式,使用Swift语言编写了昆虫图片的获取、显示、识别等功能。通过真机实时性能测试表明该方法具有良好的可靠性,满足了智能化的实时性要求。通过昆虫图像识别效果评估,结果表明该方法能够智能识别湿地微型昆虫,其识别精确度和回调率达到了92%,准确率达到了91%。
Insects in the wetlands of Erhai Lake are small and difficult to identify.In order to improve the efficiency of intelligent insect data processing and classification,an intelligent insect identification method was designed for the Erhai Lake wetland insects.The method realizes insect image classification learning through SVM-AdaBoost machine learning model.Based on IOS mobile platform,this paper adopts MVC(Model View Controller)design pattern and uses Swift language to realize the functions of insect image acquisition,display,recognition and so on.The real-time performance test shows that the method has good reliability and meets the real-time requirement of intelligence.Through the evaluation of insect image recognition,the result shows that the algorithm can recognize wetland micro-insects intelligently.Its recognition accuracy and callback rate reached 92%,and the accuracy rate reached 91%.
作者
罗桂兰
王熙
郝鸿俊
张梅
潘小雄
Luo Guilan;Wang Xi;Hao Hongjun;Zhang Mei;Pan Xiaoxiong(College of Mathematics and Computer,Dali University,Dali,Yunnan 671003,China)
出处
《大理大学学报》
CAS
2020年第6期7-13,共7页
Journal of Dali University
基金
国家自然科学基金项目(61661001)
云南省地方本科高校(部分)基础研究联合专项资金项目(2018FH001-057)
国家级大学生创新创业训练计划项目(201810679038)。
关键词
昆虫图像处理
智能分类识别
支持向量机
ADABOOST算法
准确率
insect image processing
intelligent classification and recognition
support vector machine
AdaBoost algorithm
accuracy
作者简介
第一作者:罗桂兰,副教授,主要从事物联网、智能生态研究。
