基于良序集和垄行结构的农机视觉导航参数提取算法

New algorithm for machine vision navigation of farm machine based on well-ordered set and crop row structure

根据田间作物垄行间杂草离散的特点,基于图像矩阵,运用像素子集的良序性,结合垄宽先验知识得到垄行轨迹中心。同时,系统选择图像的绿色成分为目标特征空间,滤掉了非绿色的背景噪声,为寻找垄行子集奠定了基础。在摄像头参数结构的可线性化映射区(图像中间约1/3区域),考虑移动平台的速度和系统图像采样间隔,在系统处理速度大于平台移动速率条件下,建立了单目视觉导航系统的动态方程。试验结果表明:航向角和位置参数平均误差分别约为1°和1 mm。该算法设计简洁,实现容易,可有效避免杂草等噪声的影响,对光照也有一定的适应性。

A basic requirement of automatic navigational system for farm machine is to recognize precisely the crop rows in various field conditions. In this paper, a new approach based on the density differences between weeds and plants was proposed to detect straight row lines, in which the subsets of the plant pixels were used to estimate the centers of crop rows. The green contents of the row images were chosen to eliminate the background soil features relatively completely compared with the Excess Green. The navigational parameters were obtained through fitting the row center dots. For nonlinear projection region L E [Height/3,2 ＊ Height/3] of the camera where projecting errors can be neglected, a navigation system equation was established at a platform speed of 1 m/s and sampling interval of 0. 1 s. Experimental results indicated that the tracking error of the navigation system was reduced with an angle error about 1° and position error about 1 mm. The system could work well under a wide range of illumination situations and was insensitive to small area weeds.