遮挡苹果目标的三点定位算法研究  

Research on Three-point Localization Algorithm for Occluded Apple Targets

作　　者：吴德刚[1] 赵利平[1] 陈乾辉[1] Wu Degang;Zhao Liping;Chen Qianhui(Shangqiu Institute of Technology,Shangqiu City,Henan Province 476000)

机构地区：[1]商丘工学院机械工程学院,河南商丘476000

出　　处：《黄河科技学院学报》2022年第8期40-45,共6页Journal of Huanghe S&T College

基　　金：商丘工学院2022科研项目(2022KYXM02)。

摘　　要：针对苹果采摘机器人不能准确定位遮挡苹果目标的问题,提出了一种基于三点定圆原理的目标定位算法。该算法利用K-means聚类方法分离出遮挡苹果的目标与背景,采用Canny算法和凸壳理论相融合的方法获得苹果遮挡部分的真实轮廓,运用三点定圆理论实现遮挡苹果整体目标的边缘定位。仿真试验结果显示,三点定圆算法的平均假阳性率为1.236%,比Hough变换算法降低11.727%,比轮廓曲率算法降低7.323%,三点定圆算法的平均重叠系数为88.921%,比Hough变换算法提高了30.938%,比轮廓曲率算法提高18.644%,表明该三点定圆算法提高了定位精度,为实现被遮挡苹果的自动采摘奠定了基础。For the problem that the apple picking robot cannot accurately locate the obscured apple target,a target localization algorithm based on the three-point fixed circle principle is proposed.Firstly,the K-means clustering algorithm is used to separate the apple target from the background.Then,the canny algorithm and convex hull theory are used to obtain the true outline of the apple.Finally,the three-point circular algorithm is used to achieve accurate positioning of the occlusion apple target.The test results show that the average false positive rate of the three-point calibration algorithm is 1.236%.It is 11.727%lower than the Hough transform algorithm.It is 7.323%lower than the contour curvature algorithm.The average overlap coefficient of the three-point calibration algorithm is 88.921%.It is 30.938%higher than the Hough transform algorithm.It is 18.644%higher than the contour curvature algorithm.It shows that the three-point circular algorithm improves the positioning accuracy and lays the foundation for the automatic picking of the covered apples.

关 键 词：遮挡苹果目标 K-MEANS算法 三点定圆原理 凸壳理论 

分 类 号：TP931.41[自动化与计算机技术]