基于立体视觉下联合收获机边缘检测技术的研究

Research on Edge Detection Technology of Combine Harvester Based on Stereo Vision

智能联合收割机是目前作物收获的发展热点与研究重点技术之一,而确定田间地块边缘、未收割作物边缘、联合收获机直线行驶和在作物地头转向是实现全田间自动收获的基础。为此,设计了一种基于立体视觉下多种作物收获边缘快速检测方法,保持作物收获边缘在检测目标区域内,并提出了一种基于HSV(色调、饱和度和值)空间扫描的动态感兴趣区域提取算法,利用Ostu算法获得作物未收获面积,实现未收获作物边缘和作物末端边缘的同时检测,并根据作物收获末端边缘自动转向到下一个收割路径。田间验证试验表明:水稻检测准确率高于98%,玉米的检测准确率高于94%,平均处理速度为49frame/ms。研究结果表明:提出的基于立体视觉下联合收获机边缘检测及精准转向系统具有较高的工作精度和效率,可为提高作物收获性能提供技术参考与借鉴。

Intelligent combine harvester is one of the development hotspots and research key technologies for crop harvesting at present,and the edge of unharvested crop edge in field plot,the combine harvester driving in a straight line and turning at the crop head are the basis for realizing automatic harvesting in the whole field.This study designs a fast detection method for multiple crop harvest edges based on stereo vision,keeps crop harvest edges within the detection target area,proposes a dynamic region of interest extraction algorithm based on HSV(hue,saturation and value)spatial scanning,obtains crop unharvested area using Ostu algorithm,realizes simultaneous detection of unharvested crop edges and crop end edges,and based on crop harvest.The end edges are automatically shifted to the next harvesting path.Based on the field validation experiments of rice and corn crops,the detection accuracy of rice is higher than 98%,and that of corn is higher than 94%,with an average processing speed of 49 frame/ms.The results of this study show that the proposed stereo vision-based combine harvester edge detection and precise steering system has high working accuracy and efficiency,and provides a technical reference for improving crop harvesting performance.The results show that the proposed stereo vision-based combine harvester edge detection and precise steering system has high working accuracy and efficiency,which can provide technical reference for improving crop harvesting performance.