免耕覆盖地秸秆行茬导航路径的图像检测  

Detection of Stubble Row and Inter-row Line for Computer Vision Guidance in No-till Field

作　　者：王晓燕[1] 陈媛[1] 陈兵旗[1] 李洪文[1] 孙浩[1] 

机构地区：[1]中国农业大学工学院,北京100083

出　　处：《农业机械学报》2009年第6期158-163,共6页Transactions of the Chinese Society for Agricultural Machinery

基　　金：农业部948重大专项(2006-G58)

摘　　要：提出了秸秆行茬视觉导航方法,引导农机具在根茬行间作业以防止堵塞,并对玉米直立根茬导航路径的检测方法进行了研究。选取了图像的彩度信息作为研究对象,通过多种自动阈值选取方法的对比试验和分析,采用迭代法对根茬图像与土壤背景进行了有效分割,并采用2次腐蚀处理去除噪声,通过膨胀处理填充目标区域孔洞;使用了过已知点的Hough变换检测出玉米直立根茬行间及行茬直线。田间试验表明,该方法能够有效地提取出秸秆行茬导航路径,计算时间不超过0.1 s。The residue blockage to the soil engaging parts is one of the key problems affecting the performances of no-till planter and other machine working in conservation tillage farming. Visual guidance system was studied to guide the machine working in between the stubble rows thus to solve the blockage problem in standing stubble conditions. The maize stubble rows were detected by image processing and the saturation of the field image was selected to segment the dry stubble and soil surface background. By comparing different methods of determining threshold, the iterative method was proved to be the most effective to segment the image automatically. The improved Hough transform was used to detect the inter-row line and the stubble row line to guide the machine. The experiments with large number of field images showed that the algorithm proposed can obtain the target navigation path rapidly within 0.1 s and accurately for most conditions of standing stubble covered field.

关 键 词：保护性耕作 农业机械 自动导航 机器视觉 图像分割 路径识别 

分 类 号：TP242.62[自动化与计算机技术—检测技术与自动化装置] S223.26[自动化与计算机技术—控制科学与工程]