采摘机器人识别抓取重叠番茄果实的方法研究  

Research on the Method of Picking up Overlapping Tomato Fruits by Picking Robot

作　　者：郗厚印 张栋 周涛 杨云霄 Xi Houyin;Zhang Dong;Zhou Tao;Yang Yunxiao(College of Information and Control Engineering,Qingdao University of Technology,Qingdao Shandong 266520,China)

机构地区：[1]青岛理工大学信息与控制工程学院,山东青岛266520

出　　处：《农机化研究》2021年第12期17-23,50,共8页Journal of Agricultural Mechanization Research

基　　金：国家自然科学基金项目(61703225)。

摘　　要：为适应农业果实采摘机器人自主采摘的需求,针对重叠番茄果实,设计了一种在不同光照条件下改进色差算子的组合处理识别并抓取重叠果实的方法。首先,利用中值滤波滤除噪声;然后,采用改进算子的OTSU分割算法和形态学处理对番茄图像进行背景分割,通过异或运算、腐蚀操作,使未覆盖果实与被覆盖果实的分离更加明显,进而将凸包算法、泛洪填充算法和霍夫直线检测相结合,提取出番茄的真实轮廓,并利用最小二乘法对其进行椭圆拟合,计算出果实质心进行立体匹配;最后,设计了适用于番茄果实采摘的原型机器人,主要由三自由度的机械臂构成,用于抓取果实。分别从不同角度对不同光照条件下重叠的番茄果实图像进行了实验,并进行了水果立体匹配实验,结果表明:该方法是有效、可行且高效的。In order to adapt to the demand of agricultural picking robot autonomous picking, an effective approach to identifying and grasping overlapping tomato fruit by the combination processing of improved chromatic aberration operator under different light conditions is presented. Firstly, the median filter is used to filter out the noise. Then the OTSU segmentation algorithm based on the improved operator and morphology is used to segment the tomato image from the background. Through XOR operation and corrosion operation, the separation of uncovered and covered fruits is more obvious. Furthermore, the convex hull algorithm is combined with flood filling algorithm and Hough line detection to extract the real contour of tomato, and the ellipse fitting is performed using the least square method to calculate the fruit centroid for stereo matching. Finally, a prototype robot suitable for tomato fruit picking is designed. The mechanical structure is mainly composed of a three-degree-of-freedom manipulator. The manipulator is kinematically analyzed and can be used to grab fruit. Experiments are performed on tomato fruit images overlapping under different lighting conditions from different angles. At the same time, fruit stereo matching experiments are performed. Experimental results show that the method is effective and feasible.

关 键 词：采摘机器人 果实识别 重叠果实 改进算子 OTSU分割 

分 类 号：S225[农业科学—农业机械化工程] TP391.4[农业科学—农业工程]