基于计算机视觉的甘蓝生长过程导航线提取算法

Navigation line extraction algorithm for cabbage in different growth stages based on computer vision

为改善甘蓝施药、收获机械视觉导航效果不稳定的现状,提出适应甘蓝施药、收获机械自动导航场景的视觉导航线提取算法,采用过红灰度法(2 R-G-B)对图像进行灰度化,大津阈值法对灰度图像进行分割,Hough变换方法确定甘蓝双侧导航线,使用扫描法确定甘蓝导航跟踪中心线;本文试验图像采集自中国农业科学院寿光蔬菜研发中心甘蓝示范基地,分别在甘蓝生产初期和结球采集,应用Matlab计算平台进行算法验证,使用本研究算法平均每张图片的左右导航线与中心导航线的规划时间为分别为0.086 s、0.42 s,角度平均差分别为5.71°、8.30°和3.26°;使用本文方法规划的导航线在精度上优于传统方法。

The automatic navigation technology on the field crops can improve the intelligence level of the application and harvesting machinery.At present,the navigation based on satellite communication is widely used.However,the method lacks flexibility,navigation methods based on computer vision is flexible.Meanwhile the plant morphology will be different in the early,middle and ceremonial stages of the crop,and using the same navigation method will affect the navigation effect;In this paper,taking cabbage as an example,a navigation line extraction algorithm adapted to different growth stages of cabbage was proposed.The algorithm grayscales the image by enhancing the R component feature(over-red feature)in the image color space.Based on the over-red feature,fitted navigation line is extracted using the scanning and the Hough transform method when cabbage is in the early and middle stage of cabbage cultivation;The experimental images of cabbage in the early growth stage and the festival are taken by the smart phones in the demonstration base of Shouguang Vegetable Research and Development Center of Chinese Academy of Agricultural Sciences.The average planning time of the left and right navigation lines and the central navigation line of each picture is 0.086 s and 0.42 s,themean angle differencos are 5.71°,8.30°and 3.26°respectively;the navigation effect show that using the algorithm which is proposed by the paper is better than the traditional methods.