Study On A Vision Based Detecting Device For Stem Parameter Of Hole Tray Seedlings

Grafted seedlings directly affect work efficiency and grafting success rate of the automatic grafting machine.Through the measurement parameters of seedling stems provide data basis for automatic matching rootstock and scion.In this paper,we carried out the method of detecting the parameters of seedling based on computer vision,and developed a seedling parameter automatic detection device,provides basic condition to detection of the growth status of the stems and automatic sorting of rootstock and scion.The main research work includes the following four aspects:1.Design of grafting detection platform.the function and structure of the platform are analyzed,the whole platform consists of image acquisition subsystem,auxiliary measurement subsystem,motion control subsystem,and research on selection of parts for each subsystem.2.Depth detection system.Through the early investigation of the size of the seedling and the calculation of the working space of the detection device,the depth detection method based on the geometric relation model of laser line height and distance is proposed.Combined with a CCD camera and laser designed the depth detection auxiliary device.After 200 groups of experiments,the difference between the system measurement results and the manual measurement is 94.5% within the allowable range of 0.1mm,the deviation of more than 0.1mm is 5.5%.3.Research on an approach for stem parameter measurement.Based on channel separation,threshold segmentation and other pretreatment for the stem image,study of the centerline of the laser line,quadratic curve fitting of centerline,the intersection of the curve and the edge of the stem was carried out.,stem is divided into two parts by stem axis,the parameters of the stem were measured separately,finally calculate the sum of the parameters.After 200 groups of experiments,the system measured values were compared with manual measurements,the maximum error is 0.241 mm,the average error is 0.151 mm.4.Research on motion control Subsystem.Mainly including kinematic derivation of manipulator and motor motion control.In order to ensure the movement accuracy of the robot terminal trajectory,use the sub-motion control method.After 100 groups of experiments,the average response time from the raspberry pi to the robot arm is 41.5 ms.