Research On Garlic Planter Directional Control System Based On Machine Vision

China ranks first in the world in terms of garlic planting area and production,and is the world’s largest garlic producer.When planting garlic,there is an agronomic requirement of "planting with buds pointing upward and planting upright".The garlic planted with buds pointing upward can not only enhance the quality and quality of the garlic and garlic,but also bring more economic benefits to farmers with high quality and high yield garlic.Planting garlic is mainly artificial sowing,which is not only a long time but also low efficiency,so realizing the mechanization planting of garlic is the key to realize the development of Chinese garlic industry.The existing garlic planters mainly adopt the purely mechanical method for the orientation of garlic species,which is limited to the relatively regular shape,the lower distribution of the center of gravity,and the long bud tip.It is not suitable for the general varieties of garlic species whose planting area accounts for about 2/3 in China.The use of machine vision technology can effectively solve the general garlic seed orientation problem,but the identification environment requirements are high,there is no garlic planter using the garlic orientation control system based on machine vision technology,it is necessary to develop a set of suitable for garlic planter orientation control system.In order to solve the application of garlic directional control system in the planter,this paper designs a set of garlic directional control system based on machine vision combined with four rows of air suction seeding device,which is suitable for garlic planter to meet the requirements of mechanical planting of general garlic seeds.In this paper,a directional control system of garlic planter is designed.The system includes four rows of garlic seed bud tip recognition and multi-motor direction control,so as to realize four rows of continuous single seed directional seeding.The directional control system can collect,process and recognize four garlic pictures with numbered sequence at the same time,and the recognition results are sent to the controller in numbered sequence to realize independent parallel direction control of each row of garlic species.In order to recognize multiple garlic images in the image recognition system at the same time,the thread pool technology of multithreading technology is applied in the algorithm.At the same time,in order to prevent the problem that the process is easy to be blocked when running the algorithm,the Epoll multiplexing method is added to the thread pool technology to solve the blocking problem.After the test verification,it takes 23.54 seconds to recognize 100 groups of pictures with four pictures as a group,and the average time of a group of pictures is 0.2354 seconds.Compared with Raspberry PI,industrial computer has larger memory,more pictures stored,and can run multithreaded algorithms.Therefore,industrial computer as an image processing unit has the advantages of strong stability,strong software and hardware compatibility,and strong expansibility.It can meet the requirements of simultaneously recognizing multiple garlic pictures and sending recognition results to the control system in the recognition sequence.A multi-motor alignment control system is designed.When the alignment system works,the controller can determine whether to send pulse signal to the stepper motor driver to control the stepper motor rotation alignment according to the recognition results sent by the industrial computer.Because the stepper motor of the steering system operates in parallel and independently,the method of parallel control of multiple motors is adopted in the multi-motor cooperative control.According to the working parameters of stepper motor direction adjustment,the hardware selection of the controller is carried out,and three selection schemes are formulated.After comparing the advantages and disadvantages of the three controllers and their practicability,the tabular control TPC20 is selected as the controller of the direction adjustment system.The communication between the industrial computer and the controller is also studied.The signal communication between the industrial computer and the controller can be realized by using the RS-485 converter,which can transmit multiple directional signals at a time,and solve the problem that the industrial computer comes with a variety of communication serial ports and the input interface of the controller can not match.The orientation rate and seeding efficiency of garlic species were used as the evaluation indexes of the system.The seeding efficiency was mainly affected by the identification time of garlic bud tips and the success rate of orientation.The factors that affect the evaluation indexes of the control system,such as the accuracy rate of bud tip identification,identification time,orientation rate and orientation rate,were analyzed experimentally.The results showed that the system basically met the requirements of mechanized planting.The garlic directional control system based on machine vision designed in this paper is combined with the garlic planter to realize the single continuous directional seeding of four rows of garlic,which improves the garlic planting efficiency to a certain extent.