采摘机器人视觉图像检测应用研究——基于篮球动作捕捉的多媒体数据库系统  

Application Research on Visual Image Detection of Picking Robot Based on Multimedia Database System of Basketball Action Capturing

作　　者：罗伟 Luo Wei(Shantou Technician College, Shantou 515041, China)

机构地区：[1]武汉学院,武汉430079

出　　处：《农机化研究》2018年第11期203-207,共5页Journal of Agricultural Mechanization Research

基　　金：湖北省教育厅科学技术研究计划指导性项目(B2016417)

摘　　要：采用篮球动作捕捉系统可以实时地对比赛视频进行分析,通过动作捕捉,得到优秀运动员的投篮动作技术特征;但是篮球比赛属于较高强度的对抗性比赛,其移动速度较高,因此捕捉系统需要有较高的精度。将高精度的动作捕捉系统应用在采摘机器人视觉系统设计中,可以有效提高采摘机器人对果实的识别精度,从而提高作业的自动化程度。为了验证方案的可行性,对采摘机器人的视觉系统进行了测试,并以夜间采摘环境为例,对采集的图像进行了平滑和增强处理,成功提取到了果实的边缘特征图像。对采摘机器人视觉系统的目标识别率和定位能力进行了测试,测试结果表明:目标识别率和定位准确率都较高,满足高精度采摘作业的需求。The basketball motion capturing system can be real-time analysis of the video game,through motion capturing. It can get shot technical characteristics of athletes,but the basketball game belongs to the high strength of antagonism,the moving speed is higher,so the capturing system needs to have high accuracy. It can effectively improve the recognition accuracy of picking robot and improve the automatic degree of operation by applying high precision motion capturing system to picking robot vision system. In order to verify the feasibility of the visual system of the picking robot,taking the night picking environment as an example,it tested the image smoothing and enhancement processing to the image edge feature extraction of fruit. The target recognition rate and positioning ability of the picking robot vision system are tested. The test results show that the target recognition rate and positioning accuracy are higher,which meets the needs of high-precision picking operations.

关 键 词：机器视觉 篮球动作捕捉 多媒体数据库 采摘作业 特征提取 

分 类 号：S225[农业科学—农业机械化工程] TP391.41[农业科学—农业工程]