中文摘要：

作物行骨架线提取的精准度是农业机械自动导航的关键,针对传统作物行骨架线提取算法存在的连通性差、无法保持单一像素以及存在冗余像素等问题,提出一种利用最大圆盘提取作物行骨架的改进算法。首先采用预处理操作对作物行图像进行二值化处理,然后对二值作物行图像进行细化,在细化过程中通过引入欧氏距离寻找其中点的方法保持骨架的连通性,同时采用追踪扫描的方法有效地对冗余像素进行剔除,提高骨架的拓扑结构能力。实验结果表明,该方法对成行作物具有很好的适应性,且提取的骨架线更准确,与传统的算法相比,采用本文算法提取的骨架拟合后的导航线误差相对较小,能满足实际应用需求。

英文摘要：

The accuracy of crop rows skeleton line extraction is the key to the automatic navigation of agricultural machinery.In the paper,an improved algorithm based on the maximum disc extraction algorithm is proposed to solve the problem of poor connectivity,single pixel and redundant pixels.Firstly,image pre-processing operations is used in order to obtain a binary image,then the binary image is thinned.In the refinement process,the connectivity of the skeleton is maintained by the introduction of Euclidean distance.At the same time,the method of eliminating redundant skeleton pixels was adopted to improve the ability of the topology of the crop rows skeleton.Experimental results show that the proposed method has good adaptability to the row crop,and the skeleton lines that are extracted is more accurate.Compared with the traditional algorithms,the error of the navigation line is relatively small by using this algorithm,which could meet the needs of the practical application.