基于边缘检测与扫描滤波的农机导航基准线提取方法

为了实时、准确地提取作物行基准线,提出了一种将边缘检测和扫描滤波(Boundary detection and scan filter, BDSF)相结合的基准线提取方法。首先对RGB颜色空间采用G-R颜色特征因子进行图像灰度化,再采用最大类间方差法（OSTU）对灰度图像进行分割,得到二值化图像,获取较好的作物信息。然后分别对图像的底端和顶端部分进行垂直投影,获取作物行的位置,形成一个包含作物行直线的条形框;在这个条形框内,再用等面积的小条形框对图像进行扫描并统计有效点的个数。最后根据扫描的结果来提取导航线。试验结果表明,对比Hough算法和最小二乘法(Least square method, LSM),BDSF算法处理一幅分辨率为640像素×480像素的图像,平均耗时为67ms,与LSM算法耗时相当,精度接近Hough算法;并且在杂草和株数稀缺情况下具有良好的适应性,能够快速准确地提取作物行基准线。

Crop Baseline Extraction Method for Off-road Vehicle Based on Boundary Detection and Scan-filter

To accurately detect the baseline of crop row in real time, a navigation line extraction method that combine boundary detection and scan-filter (BDSF) was proposed. Firstly, the binary image that contained the crop row information was obtained correctly after image preprocessing, including gray scale processing which adopted the G-R color feather method, and binary processing which use the OSTU method. Secondly, the position of crop row was calculated by vertical projection of the image top and bottom. Thirdly, numbers of rectangle bars of width equity was used to scan the effective information of image points in the rectangle boxes. Finally, the navigation line was extracted according to the scanning results. The experimental results showed that, the BDSF algorithm could quickly and accurately extract the navigation line. The time consumption for single image of 640 pixels×480 pixels was about 67ms. Compared with Hough transform and LSM algorithm, the designed algorithm had the advantages of high speed and good robustness.