【摘要】 为解决农业机器人单目视觉路径识别率低下、易受光照以及杂草影响的问题，提出一种适用于种植初期作物行导航线快速提取的方法。首先对图像进行归一化处理，采用改进的超绿算法(1.88g-r-b)进行灰度化，采用固定阈值法和Otsu法结合对图像进行二值化，通过中值滤波以及形态学滤波对得到的图像进行处理，设置ROI区域，消除形态学影响。利用垂直投影法对作物行特征点进行提取，提取后的特征点利用最小二乘法进行导航线拟合。试验数据表明：该算法识别效果好，精度高，实时性强，可为农业视觉导航提供依据。更多还原

【Abstract】 In order to solve the problems of low monocular vision path recognition rate of agricultural robots and easy to be affected by light and weeds, a method suitable for rapid extraction of crop guidance routes in the early stage of planting is proposed. First, normalize the image, use the improved super green algorithm(1.88g-r-b) for graying, use the combination of fixed threshold method and Otsu method to binarize the image, process the obtained image through median filter and morphological filter, set ROI area, and eliminate the influence of morphology. The vertical projection method is used to extract the feature points of crop rows, and the extracted feature points are fitted with the navigation line by the least square method. Experimental data show that the algorithm has good recognition effect, high accuracy and strong timeliness, and can provide a basis for agricultural visual navigation.更多还原