摘要
为实现对进口苹果果腐病的快速筛查与监测,本文建立了一种基于可见-近红外漫反射光谱技术的苹果果腐病快速、无损检测技术。使用功率为10 W的卤钨灯为光源,采用光纤结合积分球获取苹果的漫反射光谱,可有效避免光源照射灼伤苹果。建立了基于卷积神经网络的苹果果腐病快速筛查模型,网络可自主学习并提取光谱特征信息,无需复杂的光谱预处理,检测时间约为0.25 s,为苹果果腐病的快速筛查与监测提供了有力的技术支持。
In order to achieve the rapid screening and monitoring of fruit rot in imported apples,a rapid and non-destructive testing technology of fruit rot based on visible-near infrared diffuse reflection spectroscopy was established.In this paper,a tungsten halogen lamp of 10 W was used as the light source,while an integrating sphere combined with optical fiber was used to collect the diffuse reflection spectrum of apples,which can effectively avoid the burning of apples by light source irradiation.Furthermore,a rapid screening model of fruit rot in apples based on convolutional neural network was established,with which spectral characteristic information can be learnt and extracted by itself without any complex spectral preprocessing.The testing time was only about 0.25 s,which provides powerful technical support for the rapid screening and monitoring of fruit rot in apples.
作者
吴向峰
刘辉军
邹哲祥
任彬烽
宋佳松
张小全
WU Xiang-Feng;LIU Hui-Jun;ZOU Zhe-Xiang;REN Bin-Feng;SONG Jia-Song;ZHANG Xiao-Quan(Technology Center of Fuzhou Customs District,Fuzhou 350001;China Jiliang University,Hangzhou 310018)
出处
《中国口岸科学技术》
2021年第8期58-64,共7页
China Port Science and Technology
基金
国家重点研发计划(2016YFF0203705-2)
海关总署科研项目(2019HK058)。
关键词
苹果
果腐病
可见-近红外光谱
卷积神经网络
apple
fruit rot
visible-near infrared spectroscopy
convolutional neural network
