基于机器视觉的自主插秧机导航信息的提取研究

The Study for the Navigation Information about Automatic Navigation of Unmanned Rice Mechanical Transplanter Based on Machine Vision

针对基于机器视觉的插秧机的自主导航,提出了一种利用秧苗行分割线作为基准线提取导航参数的算法。根据秧苗在田间环境的特征,用EXG因子分割图像,将按列累加的灰度值所形成的曲线图,设定水平分割线找到秧苗区域,确定各苗区的起始列点和终止列点,找到定位点,拟合分割线。根据秧苗行分布呈平行线状的特点,利用分割线在图像平面上形成的灭点和成像的斜率来计算插秧机的位移偏差和角度偏差,为视觉导航提供必要的参数。实验结果表明该方法具有一定的可行性,能够用于插秧机器人的导航研究。

For the automatic navigation of unmanned rice mechanical transplanter based on machine vision, the paper proposed a algorithm of getting navigation parameters that used the seeding row line as baseline. For the feature of seedling in the paddy field, the seedling images were segmented by the EXG factors, the curves resulting from vertical projection of grey value of images which was found through the horizontal separating lines. The position of starting row points and ending row points in all seedling fields could be calculated and the orientation of points could be detected. The separating lines of seedling rows were accomplished by robust regression. Because the seedling rows presented parallel lines, the paper provided the vanish point of the projected lines on the image of a group of separating lines at infinity and the gradient of baselines on the image to calculate the pose of transplanter. Experimental results demonstrated that the effectiveness and the robustness of our approach, which can be applied of the automatic navigation of unmanned rice mechanical.