基于机器视觉的茶陇识别与采茶机导航方法

Tea Ridge Identification and Navigation Method for Tea-plucking Machine Based on Machine Vision

为提高机器采摘茶叶整体质量及采摘效率,提出一种基于机器视觉的机采茶陇识别与采茶机导航的方法。在采茶机器的前方配置摄像头用于获取待采摘茶陇的视频图像,对安装在采茶机上的摄像机内外参数进行离线标定;对茶陇的视频图像用高斯滤波、颜色坐标系转换、局部OTSU多阈值法图像分割、最小二乘法直线拟合等算法进行处理,提取出茶陇的左右边缘线;根据标定结果计算出茶陇中心线,并在驾驶座的显示屏上标识出目前采茶机的采摘状态,包括茶陇中心线和采茶机的偏移情况,提示驾驶员根据识别结果进行适当调整。实验结果表明,提出的基于机器视觉的机采茶陇识别及采茶机导航的方法能有效解决目前机采茶叶老嫩茶叶一刀切下的弊端,为完全自动化采摘奠定了基础。

In order to improve tea quality and efficiency of plucking by using the tea-plucking machine,this paper proposed a kind of identification method of the tea ridge and navigation method of the teaplucking machine. Firstly,the inside and outside parameters of camera offline were calibrated,which were put in the front of tea-plucking machine for getting the video information about the tea ridge.Secondly,the Gauss filter method was used to reduce noise jamming, and after customizing the rectangular window,the local OTSU thresholds method was used to segment image,the image was obtained by the camera. Then the points which belonged to the edge lines of the tea ridge were found and noted based on space constraint,the space constraint can improve the speed and the accuracy to find the points. Least square linear fitting method was used to fit the left and right edge lines of the tea ridge.Then the center line of the ridge was calculated according to the calibration results,and the state of the tea-plucking machine,including the center line of the ridge and the offset of the tea-plucking machine,was showed on the screen of the driver＇s seat in order to prompt the driver how to adjust the machine.Experimental results show that this method can identify the tea ridge accurately and navigate the teaplucking machine,and also can solve the disadvantages of cutting leaves with too many old leaves,and lay a solid foundation for fully auto-cutting type tea-plucking machine in the future.