Abstract: Aiming at the needs of automatic grading and transplanting of plug seedlings in factory production, this paper designs a set of plug seedling quality classification system based on machine vision and image processing. Firstly, the hardware system of conveying mechanism, control mechanism, transmission mechanism, actuator and machine vision system is designed.Secondly, the image processing method based on template matching is used to find plug seedlings in the front view and extract their plant height and ground diameter. And, the image segmentation method based on color information is used to find plug seedlings in the top view and extract their leaves.The height, ground diameter and leaf area of plug tray seedlings are used as feature parameters to train the SVM grading model. A polynomial kernel function SVM classifier is used to automatically classify plug tray seedlings.We have developed system control and analysis software based on Opencv3.6, Python3.7 and Qt5, which can visually edit key parameters.Based on the research and development of the plug tray seedling grading transplanting prototype, we conducted automatic grading and transplanting experiments on tomato seedlings of 7-10 days.The test results show that when the average transplanting speed is 9.6 plants/min, the transplanting success rate is 96.88%, and the classification correct rate is 95.83%.Theoretically, using 4 robots for transplanting can achieve a transplanting speed of 38 plants/min, which can work continuously and steadily, effectively solve the labor-consuming and labor-intensive problem, and improve the accuracy of seedling transplanting.The advantages of automatic grading and transplanting are more obvious for long working hours.