基于图像处理技术的蚕座中家蚕识别研究  

Study on a Recognition Method for Silkworm Seat Based on Image Processing Technology

作　　者：杨朔 左欣 赵子牛 熊孟 张国政 颜辉[1,2] Yang Shuo;Zuo Xin;Zhao Ziniu;Xiong Meng;Zhang Guozheng;Yan Hui(School of Biotechnology,Jiangsu University of Science and Technology,Zhenjiang Jiangsu 212100,China;Sericultural Research Institute,Chinese Academy of Agricultural Sciences,Zhenjiang Jiangsu 212100,China;School of Computer,Jiangsu University of Science and Technology,Zhenjiang Jiangsu 212100,China)

机构地区：[1]江苏科技大学生物技术学院,江苏镇江212100 [2]中国农业科学院蚕业研究所,江苏镇江212100 [3]江苏科技大学计算机学院,江苏镇江212100

出　　处：《蚕业科学》2022年第5期430-436,共7页ACTA SERICOLOGICA SINICA

基　　金：江苏省现代农业重点研发项目(BE2020331);财政部和农业农村部国家现代农业产业技术体系项目。

摘　　要：家蚕与蚕座中桑叶(包括桑枝)背景的准确分割技术,是基于数字图像处理技术开发家蚕自动饲喂系统的关键技术,也是开发智能化、自动化养蚕系统的重要组成部分。试验以5龄家蚕(菁松)为研究对象,饲喂条桑,采用摄像机连续录制蚕座图像。首先采用Otus法与最大熵法对家蚕和桑枝(包括叶脉)图像进行处理,然后对图像RGB各分量数值进行分类统计分析,最终获得基于B分量的区分家蚕与桑枝的分割阈值。图像经此分割后,计算家蚕在蚕座中的面积占比。结果表明,蚕座图像经过Otus法与最大熵法处理去除了桑叶图像,在B分量上获得了最佳分割阈值129,去除了桑枝(包括叶脉),最终从图像中分割出家蚕,建立了家蚕在蚕座中面积占比的实时计算方法。经验证,该方法准确可靠,为开发自动化的家蚕自动饲喂系统提供了实验依据。The accurate segmentation of the silkworms and the background mulberry leaves(including mulberry branches) in the silkworm seat is the key technology for the development of an automatic silkworm feeding system based on image processing technology, and is an important part of the development of an intelligent and automated silkworm rearing system. The 5 th-instar silkworm(Jingsong), used as the research object, was fed with the mulberry branches. Meanwhile, a video camera continuously recorded the silkworm seat. The images of the silkworms and mulberry branches(including leaf veins) were first processed by the Otus method and the maximum entropy method, then the RGB component values were classified and statistically analyzed, and finally a segmentation threshold for distinguishing silkworms and mulberry branches based on the B component was obtained. After the image segmentation, the area proportion of silkworms in the silkworm seat was calculated. The results showed that the Otus method and the maximum entropy method removed the mulberry leaf images, and the best segmentation threshold of 129 was obtained on the B component, which realized the removal of mulberry branches(including leaf veins), and the segmentation of silkworms from the image. Accordingly, a real-time calculation method for the area of silkworms in the silkworm seat was established. It has been verified that the method was accurate and reliable, providing an experimental basis for the development of an automated silkworm feeding system.

关 键 词：图像识别 蚕座 RGB颜色空间 图像阈值 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] S883[自动化与计算机技术—计算机科学与技术]