摘要
在猕猴桃采摘机器人作业的过程中,由于对于目标果实的识别一般是针对单个果实,极大影响了猕猴桃采摘的效率。为此,提出多目标果实的识别方法,以猕猴桃果萼为识别对象对猕猴桃图像进行识别。运用该方法对多目标果实进行识别,对所采集的样本图像中的20幅图片进行实验,实验结果表明,运用该方法识别的准确率为88.86%,该识别获得的信息能达到猕猴桃多目标识别的目的,为后续机械臂的路径规划,采摘机器人末端执行器设计提供了信息支持,也可为苹果、柑橘等果实的识别提供参考。
When kiwifruit harvesting robot is working,the recognition of the target fruit is aiming at a single one generally,which affects the picking efficiency of robot greatly.For this issue,a recognition method of multi-target was proposed,which recognized kiwifruits images by referring the calyx as the subject.20 images of the sample images were tested using the proposed recognition method of multi-target,the conclusion was drawn that the accuracy of the method can reach 88.86%.The information obtained using the method achieves the goal of multi-target recognition of kiwifruits successfully and it can provide information supports for the following path planning of mechanical arms and designing of end-effector.The method also can provide references for recognition of other fruits,such as apples,oranges,etc.
作者
陈礼鹏
穆龙涛
刘浩洲
谷新运
傅隆生
崔永杰
CHEN Li-peng, MU Long-tao, LIU Hao-zhou, GU Xin-yun, FU Long-sheng, CUI Yong-jie(College of Mechanical and Electronic Engineering, Northwest Agricultural and Forestry University, Yangling 712100, Chin)
出处
《计算机工程与设计》
北大核心
2018年第6期1738-1744,共7页
Computer Engineering and Design
基金
国家自然科学基金项目(61175099)
陕西省科技统筹创新工程计划基金项目(K3320215126)
陕西省果业专项基金项目(2013159)
关键词
采摘机器人
多目标
果实识别
猕猴桃果萼
图像处理
picking robot
multi-target
fruit recognition
kiwifruit calyx
image processing
