基于FCM及HSV模型的方格蔟黄斑茧检测与剔除技术  

Detection and Elimination of Yellow Spotted Cocoon in Mountage Based on FCM Algorithm and HSV Color Model

作　　者：刘莫尘[1,2] 许荣浩 闫筱 闫银发 李法德[1,2] 刘双喜 LIU Mochen;XU Ronghao;YAN Xiao;YAN Yinfa;LI Fade;LIU Shuangxi(College of Mechanical and Electronic Engineering, Shandong Agricultural University, Taian 271018, China;Shandong Provincial Key Laboratory of Horticultural Machinery and Equipment, Taian 271018, China;School of Electrical and Electronic Engineering, North China Electric Power University （Baoding） , Baoding 071003, China)

机构地区：[1]山东农业大学机械与电子工程学院,泰安271018 [2]山东省园艺机械与装备重点实验室,泰安271018 [3]华北电力大学(保定)电气与电子工程学院,保定071003

出　　处：《农业机械学报》2018年第7期31-38,共8页Transactions of the Chinese Society for Agricultural Machinery

基　　金：国家现代农业产业技术体系建设专项(CARS-18-ZJ0402);山东省现代农业产业技术体系建设项目(SDAIT-18-06);山东省"双一流"奖补资金项目(564047)

摘　　要：桑蚕业缫丝前需要对蚕茧进行检测分类,剔除黄斑茧等下茧对于提高缫丝质量非常关键。我国蚕虫上蔟多用纸板方格蔟,方格蔟在使用过程由于挤压、受潮等原因极易变形,另外,营茧方格蔟一般覆盖一层茧衣,使自动检测方格蔟蚕茧质量并准确剔除黄斑茧等下茧极其困难。采用基于模糊C均值聚类(FCM)及HSV模型的黄斑茧检测算法,在采茧过程中直接对蚕茧进行图像分割和黄斑茧检测,使用基于机器视觉的直角坐标式自动采茧机对黄斑茧进行定位剔除试验。首先对方格蔟正面原始图像使用FCM分割,消除蚕茧茧衣及方格蔟边框,得到蚕茧二值图像;对FCM分割后的蚕茧二值图像与方格蔟原始图像进行掩膜,实现对方格蔟内单个蚕茧的提取;对提取到的单个蚕茧,根据HSV空间累积颜色直方图的黄斑颜色H分量的比例是否达到黄斑茧的定义阈值,逐个进行黄斑茧判断;然后,对检测到的黄斑茧,保存其连通域中心坐标作为其图像位置坐标,经过视觉测量确定蚕茧在笛卡尔空间中的世界坐标。使用基于机器视觉的直角坐标式方格蔟自动采茧机,对方格蔟黄斑茧进行定位剔除试验,该算法对方格蔟内黄斑茧的平均检测正确率为81.2%,黄斑茧坐标最大定位偏差为3.0 mm,对单张方格蔟图像进行分割和黄斑茧检测的平均时长为1.271 s,对茧衣附有桑叶梗或碎桑叶的蚕茧没有误检测,但对黄斑位于边缘处的蚕茧检测效果不好。In sericulture, cocoons must be detected and classified before silk reeling is performed. It is important for improving the quality of silk to eliminate the yellow spotted cocoons. A large number of checker cocooning are used for silkworm mounting cocooning frames. It is difficult to detect the yellow spotted cocoons because of the checker cocooning frames’ reshape and outer floss. To solve the problem, the algorithm based on FCM and HSV color model was used to detect and eliminate the yellow spotted cocoons in the cocoons harvested process. Firstly, FCM segmentation was applied to the original image of the checker cocooning frame to eliminate the outer floss and the frame. The binary image of the cocoon was obtained by FCM segmentation and threshold segmentation. The original image was masked with the binary image which was obtained by FCM segmentation. And the individual cocoon was extracted through the masked operation. According to the proportion of specific color components in the color histogram which was gotten by accumulating color of HSV, the yellow spotted cocoon was judged one by one. The center point coordinates of the yellow spotted cocoons’ regions were got by the connected components calibration, and were mapped into the world coordinates through the equation that image coordinates to world coordinates to get the cocoons positions in the Cartesian space. Finally, the yellow spotted cocoons were eliminated by automatic harvesting machine. According to the result of experiment, the correct ratio of cocoon detection was 81.2%, the location accuracy was 3.0mm, the average process time of one mountage image was 1.271s. The cocoons which out floss with leaf stalks or crushed leaves could be detected errorlessly, but the algorithm had no effect on detection the cocoons with stained point in the edge.

关 键 词：蚕茧 方格蔟 黄斑茧检测 模糊C均值聚类 颜色直方图 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] S126[自动化与计算机技术—计算机科学与技术]