Design And Testing Of A Machine Vision-Based Garlic Seed Identification And Clamping Test Bench

As an important food crop in China,garlic ranks among the top in the world in terms of planting area and production,and mechanized and intelligent planting methods have become an inevitable trend.Directional sowing has a positive impact on the seedling rate,growth state and yield of garlic.The existing garlic directional sowing machines are all mechanical in structure,which are prone to misdirection due to irregular shape of garlic seeds.In this study,a machine vision-based garlic seed identification and clamping method is proposed to address the problem of garlic seed orientation,and a method is designed to identify the position and orientation of garlic seeds using the differences in morphological and textural characteristics of garlic seeds.The robot arm and pneumatic jaws are used to pick the garlic seeds,and the trajectory presetting function of the robot arm is used to put the garlic seeds into the designated position according to the uniform orientation.The details of this paper are as follows:(1)Design and implementation of the garlic seed identification and clamping test bench.According to the design of identification clamping and weaving test bench,the functions of each component of the test bench are introduced in detail,and the main vibrating feeding device,dark box,transmission device,control cabinet,mechanical arm,clamping jaws and other parts are composed to determine the parameters of each component.The vibrating seed feeding tank is used for feeding garlic seeds into the conveyor belt,so that the seeds are evenly and orderly arranged,and the designed transverse spacing is 20 mm,the vibration force is about 3 N,and the garlic seeds are fed at an interval of 4 cm.The transverse conveyor belt is designed for garlic seed transfer,identification and clamping operation,with blue color and adjustable speed of the conveyor belt.(2)Analysis and research of pneumatic flexible clamping jaws and mechanical arms.Analyze the role of mechanical arm and jaws,explain the mechanical structure of mechanical arm and jaws,discuss the composition,working principle,parameters of jaw arm and setting of positive and negative pressure,clamping and placing process of pneumatic type jaws,determine the working positive and negative pressure of jaws as 100 KPa and 0Pa,the elastic deformation of jaw arm is about 3mm when clamping garlic seeds,and the clamping arm provides holding force of 1.6N.Analyze the composition,working The principle,working principle,control principle,determine the position of each part through the world coordinate system I,analyze the dynamic tracking function of the robot arm and the basic logic of clamping garlic seeds,analyze the principle of garlic seed orientation placement,and determine the motion interval and coordinate interval of the robot arm.(3)Research of garlic seed identification method.The image recognition process of machine vision system and garlic seed is described;the introduction of garlic seed image acquisition and pre-processing process,including image grayscale,sharpening enhancement,denoising corrosion,linear grayscale transformation and image segmentation,binarization processing,contour extraction,and determining the threshold of garlic seed recognition as 152;the feature extraction method of garlic seed image is proposed,and the area,perimeter,center of gravity,complexity,deep length,long and short axis 6 The morphological features of garlic species were extracted,and the morphological features of garlic species were obtained in the range of 24000～36000 pixels area,1000～1800 perimeter,less than 90 complexity,and 1.3～1.8 long-short axis ratio;the MATLAB machine learning model for garlic species recognition was constructed,and the varieties of garlic species could be judged based on the feature values of garlic species;the grayscale co-occurrence matrix was used to determine the energy,entropy,moment of inertia,local smoothness,and correlation 5 texture features are extracted to calculate the orientation of the garlic seed region of the image;illustrate the garlic seed recognition method and the functions of the garlic seed recognition software,set the morphological features and texture feature values of the garlic seed image,and construct the garlic seed recognition model;conduct recognition tests for the recognition model and software,and recognize 600 garlic seed images with an average recognition rate of 94% and a recognition processing time of 0.3s for each image.(4)Conduct tests on the test bench to determine the range of test factors and the maximum clamping rate.With the clamping success rate of the mechanical arm and clamping jaws as the test index,the speed of the conveyor belt,the projection size of the garlic seeds on the plane of the conveyor belt,and the height of the clamping jaws when clamping the garlic seeds as the test factors,the overall performance of the test bench was analyzed and determined using the single-factor test and the three-factor three-level orthogonal test.The single-factor test was conducted to determine the trends and causes of the variation between each factor and the clamping rate of garlic seeds,and the clamping rate of the test bench for garlic seeds was about 90%.According to the three-factor,three-level orthogonal test,the optimal parameters were determined as 0.11m/s of conveyor speed,510mm2～540mm2 of garlic seed plane size,and 3.5mm of clamping height,and the predicted clamping success rate was 93.73%.The clamping rate was greater than 90% for all three varieties of garlic.