基于有序轮廓链的葫芦科砧木苗生长点检测  

Detection of Growth Point on Cucurbitaceous Rootstock Seedlings Based on Ordered Contour Chain

作　　者：查杨 陈武 赖一波 吴尧 方家吉 黄铮 喻擎苍 ZHA Yang;CHEN Wu;LAI Yibo;WU Yao;FANG Jiaji;HUANG Zheng;YU Qingcang(Zhejiang Sci-tech University,Hangzhou,Zhejiang 310018)

机构地区：[1]浙江理工大学,浙江杭州310018

出　　处：《热带农业工程》2022年第6期28-36,共9页Tropical Agricultural Engineering

基　　金：国家自然科学基金(No.5137406)。

摘　　要：为更好地提高葫芦科幼苗嫁接设备的自动化程度,解决葫芦科幼苗插接法嫁接过程中生长点定位的问题,实现葫芦科幼苗插接法的全自动化,本文提出了基于有序轮廓链的葫芦科砧木苗生长点检测方法,通过目标像素的有序轮廓链,获取嫁接苗边缘信息,提取边缘曲率特征,或是提取目标区域单像素骨架,并以此为基础实现嫁接苗生长点定位。结果显示,俯视图的生长点检测其误差最小为1个像素值,最大为5个像素值,而侧视图的生长点定位结果误差最小为1个像素值,最大为4个像素值,因此,本算法可以准确的定位砧木苗生长点坐标并有效避免葫芦科幼苗嫁接过程中因生长点坐标偏差导致的嫁接失败的问题,大幅度提高嫁接成功率。In order to improve the degree of automation of the cucurbit seedling grafting equipment, solve the problem of locating the growing point during the grafting process of cucurbita seedling plugging method and accomplish the full automation of cucurbitaceous seedling plugging method, this paper proposes a method for detecting of growth point on cucurbitaceous rootstock seedlings based on ordered contour chain,which obtains the edge information of the grafted seedling and then extracts the edge curvature feature or the single-pixel skeleton of the target area through the ordered contour chain of the target pixel. Based on this method, the growth point of the grafted seedlings is located. The results show that the error of the growth point detection in the top view has a minimum error of 1 pixel and a maximum error of 5 pixels,while the error of the growing point location in the side view is the minimum value of 1 pixel and the maximum value of 4 pixels. Therefore, the algorithm can accurately locate the growth point coordinates of rootstock seedlings and effectively avoid the problem of grafting failure caused by the deviation of the growing point coordinates during the grafting process of cucurbitaceae seedlings, and significantly improve the success of grafting.

关 键 词：机器视觉 自动嫁接 生长点提取 轮廓检测 

分 类 号：S224[农业科学—农业机械化工程]