基于机器视觉的耕作机器人行走目标直线检测

Fast Detection of Furrows Based on Machine Vision on Autonomous Mobile Robot

针对农田耕作机器人,提出了基于机器视觉的机器人行走目标——犁沟线斜率的检测算法。将摄像机安装在拖拉机前方,在耕作过程中采集农田场景图像。根据已耕作区域、未耕作区域和非农田区域的不同颜色特征,判断出田端和犁沟线的位置以及计算斜率用的方向候补点群,使用基于一点的改进哈夫变换算法计算出犁沟线的斜率。经过对多幅实际耕作现场图片的处理,验证了本犁沟线检测算法具有速度快、抗干扰、准确性高等优点。

Machine vision navigation is indispensable to the realization of agricultural sustainable development. And the detection of the margin lines by using vision system is the first step of vision navigation. For the complexity of the farmland, it＇s very difficult to detect the margin lines accurately by using common edge-detection arithmetic, such as Roberts, Prewitt and etc. The problem could be solved by the method of improved Hough transform effectively. First, pictures were captured with the camera that was bundled in front of the tractor, then the end of the farmland and the suspected furrow were detected, at last the slope of the furrow was identified. Compared with the traditional Hough transform, the algorithm mentioned above has its own characteristics. It is high speed, no need for big memory space and with high veracity.