一种图像识别技术下的果实成熟度应用研究  

Application of Fruit Maturity Based on Image Recognition Technology

作　　者：万小红 郝蕊洁 Wan Xiaohong;Hao Ruijie(College of Mathematics and Information Technology,Yuncheng University,Yuncheng 044000,China)

机构地区：[1]运城学院数学与信息技术学院,山西运城044000

出　　处：《农机化研究》2024年第1期207-211,共5页Journal of Agricultural Mechanization Research

基　　金：山西省教育科学“十四五”规划2021年度规划课题(GH-21060)。

摘　　要：以自然环境下的红提葡萄果穗为研究对象,利用图像识别技术对其果穗成熟度进行研究分析。首先,采用Faster R-CNN卷积神经网络模型识别红提葡萄图像中的果穗,再利用KNN算法分割红提葡萄图像中的果穗和背景,并借助圆形Hough变换法检测出果穗图像中的红提葡萄果粒;最后,结合HSV空间中的H值将果粒成熟度划分为4个等级,并计算各个等级的果粒数量占果穗总果粒数量的比重,以此判断该果穗整体的成熟度,从而确定其是否能够满足采摘的要求。实验结果表明:该方法判断果穗成熟度的准确率能达到90%,满足红提葡萄果穗成熟度判断的需求,可辅助精准采摘作业。In this paper,the ear maturity of red grape in natural environment is studied and analyzed by using image recognition technology.Firstly,the Faster R-CNN convolution neural network model is used to identify the ears in the red grape image,then the KNN algorithm is used to segment the ears and background in the red grape image,and then the red grape seeds in the ear image are detected by the circular Hough transform method.Finally,combined with the H value in HSV space,the fruit maturity is divided into four grades,and calculate the proportion of the number of fruit grains of each grade in the total number of fruit grains of the ear,so as to judge the overall maturity of the ear and determine whether it can meet the requirements of picking.The experimental results show that the accuracy of this method to judge the ear maturity can reach 90%,indicating that this method can meet the needs of judging the ear maturity of red grape and realize accurate picking.

关 键 词：果穗成熟度 图像识别 Faster R-CNN KNN算法 Hough变换法 

分 类 号：S126[农业科学—农业基础科学] TP391.41[自动化与计算机技术—计算机应用技术]