Localization of Agricultural Vehicle Using Landmarks Based on Omni-directional Vision
Localization of Agricultural Vehicle Using Landmarks Based on Omni-directional Vision
复制标题
基于全方位视觉的农用车地标定位
DOI：
--
复制
发表时间：
--
期刊：
影响因子：
--
通讯作者：
S. Yokoyama
中科院分区：
文献类型：
--
作者：
M. Li;K. Imou;S. Tani;S. Yokoyama
本文介绍了一种基于全方位视觉的农用车辆路标定位系统。该方法在一块农田的一角设置四个红色的人工地标，使其成为矩形.该算法提取红色像素点作为小区域，计算平均点作为界标，利用全方位图像估计四个角度测量值。根据四个标志点之间四个夹角的几何变换，利用四条弧的四个交点的重心来估计车辆的位置。实验结果表明，该方法的最大误差和平均误差均低于普通GPS，跟踪效果较好。该算法和方法是可行的，可以替代GPS，满足未来农用车辆导航系统的需要。
This paper describes a system would be for localization of agricultural vehicle in farm fields using landmarks based on Omni-directional vision. The approach set four red artificial landmarks as rectangle in the corner of one farm field. The algorithm extracted red pixel points as a small area, calculated average point as landmarks and estimated four angle measurements using Omni-directional image. Location of vehicle was estimated by the center of gravity of four intersections of four arcs according to geometric transformation of four angles among four landmarks. Experimental results show that the maximum and mean errors were lower than common GPS and the tracking result is better. In conclusion, the algorithm and the method are feasible and could substitute for GPS to satisfy for navigation system of agricultural vehicle in future.