基于双目视觉的葡萄采摘防碰路径规划方法研究  

A study on Anti-Collision Path Planning Method of Grape Picking Based on Binocular Vision

作　　者：马聪 Ma Cong(Institute of Agricultural Economics and Information Technology,Ningxia Academy of Agriculture and Forestry Sciences,Yinchuan,Ningxia 750002)

机构地区：[1]宁夏农林科学院农业经济与信息技术研究所,宁夏银川750002

出　　处：《宁夏农林科技》2024年第8期68-71,共4页Journal of Ningxia Agriculture and Forestry Science and Technology

基　　金：宁夏自然科学基金项目“基于机器视觉的葡萄免碰采摘路径规划方法研究”(2023AAC03408);宁夏农林科学院科技创新引导项目“宁夏酿酒葡萄智慧种植关键技术研究与示范”(NKYG-23-02)。

摘　　要：目前投入实际生产的葡萄机械化采收和人工采收方式存在适应性不强、成本高、劳动力缺口大等问题,需要研发拟人化作业的智能采摘设备以提升效率,路径智能规划是实现葡萄的无损采摘的重要研究内容。以宁夏棚架式栽培的葡萄串为目标,基于双目立体视觉开展葡萄防碰采摘路径规划方法研究。采用极线校正方法进行葡萄双目图像校正,基于多种颜色空间模型结合的方法分割出左目图像葡萄串目标区域,通过局部立体匹配算法计算左目分割图像与右目原图视差值,求解出葡萄串区域范围内所有像素点的三维坐标,形成免碰路径。结果表明,采用的研究方法数据处理量小、计算结果准确,可为研发葡萄智能采摘设备提供理论基础。At present,the mechanized and artificial grape harvesting methods put into actual production have some problems such as poor adaptability,high cost,and large labor gap.It is necessary to develop intelligent picking equipment with anthropomorphic operation to improve efficiency.Intelligent path planning is an important research content for realizing non-destructive grape picking.Based on binocular stereo vision,the grape anti-collision picking path planning method was studied for the trellised grape clusters in Ningxia.The binocular grape image was corrected by the polar line correction method,and the grape cluster target area of the left eye image was divided based on the combination of multiple color space models.The parallax value between the left eye segmentation image and the right eye original image was calculated by the local stereo matching algorithm,and the three-dimensional coordinates of all pixels within the grape cluster area were solved to form a collision free path.The results show that the proposed method has small data processing capacity and accurate calculation results,which can provide a theoretical basis for the development of intelligent grape picking equipment.

关 键 词：双目视觉 葡萄采摘 防碰 空间坐标 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP242[自动化与计算机技术—计算机科学与技术]