| 基于线结构光视觉的番茄重叠果实识别定位方法研究 |
| |
| 引用本文： | 冯青春,程伟,杨庆华,荀一,王秀.基于线结构光视觉的番茄重叠果实识别定位方法研究[J].中国农业大学学报,2015,20(4):100-106. |
| |
| 作者姓名： | 冯青春 程伟 杨庆华 荀一 王秀 |
| |
| 作者单位： | 北京农业智能装备技术研究中心/国家农业智能装备工程研究中心/农业部农业信息技术重点实验室/农业智能装备技术北京市重点实验室, 北京 100097;北京农业智能装备技术研究中心/国家农业智能装备工程研究中心/农业部农业信息技术重点实验室/农业智能装备技术北京市重点实验室, 北京 100097;浙江工业大学 机械工程学院/特种装备制造与先进加工技术教育部重点实验室, 杭州 310032;浙江工业大学 机械工程学院/特种装备制造与先进加工技术教育部重点实验室, 杭州 310032;浙江工业大学 机械工程学院/特种装备制造与先进加工技术教育部重点实验室, 杭州 310032;北京农业智能装备技术研究中心/国家农业智能装备工程研究中心/农业部农业信息技术重点实验室/农业智能装备技术北京市重点实验室, 北京 100097 |
| |
| 基金项目： | 国家‘863''计划资助项目(2012AA101903);果类蔬菜产业体系北京市创新团队项目 |
| |
| 摘 要： | 为实现番茄自动化采收,针对温室光照条件多变、背景复杂及果实簇生遮挡的特征,构建了基于线结构光视觉的番茄果实识别定位视觉系统,并提出了一种针对簇生重叠果实的区域分割方法。在(2R-G-B)色差模型下利用动态阈值分割,提取成熟果实区域;计算亮度渐变梯度向量,采用灰度重心法提取结构光条纹中心线;基于线结构光立体测量原理,在光条纹扫描视场过程中实时获取成熟果实表面三维信息,并映射形成果实表面与相机光心距离深度图像,实现簇生重叠果实区域的立体分割;基于果实表面激光条纹点空间坐标拟合果实外接球体,获取果实形态和空间坐标信息;试验结果表明,视觉系统单次识别测量耗时2.8s,果实直径测量误差小于6mm,果实与摄像机中心距测量误差小于8mm。
|
| 关 键 词： | 番茄采摘机器人 线结构光 立体视觉 果实遮挡 |
| 收稿时间： | 2014/11/10 0:00:00 |
Identification and localization of overlapping tomatoes based on linear structured light vision system |
| |
| FENG Qing-chun,CHENG Wei,YANG Qing-hu,XUN Yi and WANG Xiu.Identification and localization of overlapping tomatoes based on linear structured light vision system[J].Journal of China Agricultural University,2015,20(4):100-106. |
| |
| Authors: | FENG Qing-chun CHENG Wei YANG Qing-hu XUN Yi and WANG Xiu |
| |
| Institution: | Beijing Research Center of Intelligent Equipment for Agriculture/National Research Center of Intelligent Equipment for Agriculture/Key Laboratory ofAgri-informatics, Ministry of Agriculture/Beijing Key Laboratory of Intelligent Equipment Technology for Agriculture, Beijing 100097, China;Beijing Research Center of Intelligent Equipment for Agriculture/National Research Center of Intelligent Equipment for Agriculture/Key Laboratory ofAgri-informatics, Ministry of Agriculture/Beijing Key Laboratory of Intelligent Equipment Technology for Agriculture, Beijing 100097, China;Key Laboratory of Special Purpose Equipment and Advanced Processing Technology of Ministry of Education, Zhejiang University of Technology, Hangzhou 310000, China;Key Laboratory of Special Purpose Equipment and Advanced Processing Technology of Ministry of Education, Zhejiang University of Technology, Hangzhou 310000, China;Key Laboratory of Special Purpose Equipment and Advanced Processing Technology of Ministry of Education, Zhejiang University of Technology, Hangzhou 310000, China;Beijing Research Center of Intelligent Equipment for Agriculture/National Research Center of Intelligent Equipment for Agriculture/Key Laboratory ofAgri-informatics, Ministry of Agriculture/Beijing Key Laboratory of Intelligent Equipment Technology for Agriculture, Beijing 100097, China |
| |
| Abstract: | According to the variable illumination conditions, complex background and overlapping targets of tomato, a new method of identification and localization was proposed based on linear structured light vision system.The ripe fruit area was firstly extracted under the dynamic threshold segmentation in (2R-G-B) model.The blue line laser stripe area was then determined in B channel.The direction for image of light stripe was combined with gray barycenter method for extracting the center of laser stripe area.The ripe fruit area was then filled with gray which maped the depth information of the fruits based on the spatial localization property of the linear structured light and the difference of depth between overlapping tomatoes.The segmentation and the spatial information of the overlapping tomato were finally obtained.The field test showed that the measurement error on the fruit radius and the centre distance error between the fruit and camera were less than 6 mm and 8 mm respectively. |
| |
| Keywords: | tomato harvesting robot linear structured light stereo vision target occlusion |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《中国农业大学学报》浏览原始摘要信息 |
| 点击此处可从《中国农业大学学报》下载免费的PDF全文 |
|
