基于视觉的茶作物行间行走路径规划研究

发布时间：2018-03-25 00:00

本文选题：图像处理　切入点：农业导航　出处：《农机化研究》2017年01期

【摘要】：以茶作物为对象,提出了一种基于视觉的作物行间行走路径规划方法。对实时彩色作物图像用2G-RB算法灰度化,分离作物和背景,通过固定阈值二值化图像,采用形态运算去除干扰。以区域内作物的比例变化为判别垄间垄头依据。对于垄间,利用漫水填充算法分离出垄沟,通过最小二乘法拟合导航线,得到垄间行走时候的导航控制参数;对于垄头,先分离出转弯一侧的作物,采用圆形模型拟合垄头处轮廓,通过拟合的圆来规划出转弯时候路径。试验结果表明:该算法能准确提取导航线,处理一帧图像约为150ms左右,可满足导航的要求。
[Abstract]:A vision-based path planning method for crop interline walking is proposed in this paper. The real-time color crop image is grayscale by 2G-RB algorithm, which separates crop from background, and binds the image by fixed threshold. Morphological operation is used to remove interference. The variation of crop proportion in the region is taken as the basis for distinguishing ridges between ridges. For ridges, the furrows are separated by overflowing water filling algorithm, and the navigation lines are fitted by the least square method. The navigation control parameters are obtained when the ridges are walking. For the ridge head, the crop on the turning side is separated first, and the contour of the ridge head is fitted with a circular model. The experimental results show that the algorithm can accurately extract the navigation line and process a frame of image about 150ms, which can meet the requirements of navigation.
【作者单位】：湖北工业大学农业机械工程研究设计院;
【基金】：国家自然科学基金项目(51174084) 武汉市科技局科技攻关项目(2015020202010129)
【分类号】：S22;TP391.41

【相似文献】