基于图像识别的小蚕饲育中蚕座位置与桑叶饲喂均匀性研究  

Study on Rearing Bed Position and Feeding Uniformity of Mulberry Leaf in Young Silkworm Rearing Based on Image Recognition

作　　者：何玉 石洪康 田涯涯 徐永健 汤自强 陈忠国 蒋猛[1] He Yu;Shi Hongkang;Tian Yaya;Xu Yongjian;Tang Ziqiang;Chen Zhongguo;Jiang Meng(College of Engineering and Technology,Southwest University,Chongqing 400715,China;Institute of Sericulture,Sichuan Academy of Agricultural Sciences,Nanchong Sichuan 637000,China;Chongqing Hongjiang Machinery Co.,Ltd.,China State Shipbuilding Co.,Ltd.,Chongqing 402160,China;Agriculture and Animal Husbandry Bureau of Bozhou District,Zunyi Guizhou 563100,China)

机构地区：[1]西南大学工程技术学院,重庆400715 [2]四川省农业科学院蚕业研究所,四川南充637000 [3]中国船舶重工集团公司重庆红江机械有限责任公司,重庆402160 [4]贵州省遵义市播州区农牧局,贵州遵义563100

出　　处：《蚕业科学》2019年第5期740-745,共6页ACTA SERICOLOGICA SINICA

基　　金：重庆市科技创新专项(No.cstc2017shms-xdny80048,cstc-2017shms-xdny80030);重庆市技术创新与应用发展专项(No.cstc2019jscx-gksbX0119);四川省创新能力提升工程青年基金专项(No.2019-10)

摘　　要：桑叶的自动饲喂技术是小蚕饲育智能饲喂机的核心技术,必须依据蚕座位置进行饲喂,并对桑叶撒喂的均匀性和饲喂量进行判定。利用计算机图像识别技术,通过建立坐标系、蚕座轮廓矩形化方法,实现蚕座位置的自动确定;采用模糊方法和网格划分原理,根据蚕座灰度图像中黑白像素点的分布状况,对桑叶撒喂均匀性进行判定;通过统计蚕座灰度图像中黑像素点总数占整幅图中黑白像素点总数的比率,实现对饲喂量的判定。对样本进行实际测试,与人工判定相比,蚕座位置识别正确率在92%以上,桑叶撒喂均匀性判定正确率在90%以上,饲喂量判定正确率在88%以上。试验结果表明,机器识别接近人工经验判断,具有良好的推广潜力,为小蚕饲育的智能化研究提供了参考。Automatic mulberry leaf feeding technology is the core technology of intelligent young silkworm feeding machine. The silkworm must be fed according to rearing bed position,and meanwhile feeding uniformity and feeding quantity of mulberry leaf must be determined. In this paper,computer image recognition technology was used to automatically recognize rearing bed position by establishing coordinate system and rectangularizing rearing bed contour. Feeding uniformity was determined in accordance to distribution of black and white pixels in grayscale image of rearing bed by fuzzy method and mesh generating. Feeding quantity was determined by calculating the rate of total number of black pixels in the total number of black and white pixels in grayscale image of rearing bed. Actual test showed that recognition accuracy of rearing bed was above 92%,judging accuracy of feeding uniformity was above 90%,and judging accuracy of feeding quantity was above 88% compared with manual judgement. These results showed that machine recognition is close to manual judgment by experience,which has great potential for popularization,and could provide a reference for research on intelligent silkworm rearing.

关 键 词：图像识别 边缘检测 蚕座 小蚕饲育 

分 类 号：S883.6[农业科学—特种经济动物饲养] TP273[农业科学—畜牧兽医]