Research On Rapid Identification And Orientation Device Of Sugarcane Bud

Sugarcane planting machinery adopts blind seeding method to plant sugarcane.It is impossible to ensure that the sugarcane buds face the seeding ditch wall when the sugarcane seeds fall.The planting agronomic level is low and the planting quality is not high,resulting in the production of mechanically planted sugarcane per mu instead of artificially planted mu With high output,the development of sugarcane planting machinery has encountered bottlenecks.Starting from the agronomic requirements of sugarcane planting,this topic proposes a method of combining computer vision and mechanical structure to realize sugarcane directional planting,and designs a sugarcane seed orientation adjustment system based on machine vision.The key link in the system is sugarcane bud visual recognition.The positioning method and the sugarcane seed orientation adjustment device are studied.In the aspect of sugarcane bud visual recognition,the traditional image recognition method and the convolutional neural network image recognition method are used to carry out the research,realize the two schemes of the sugarcane bud recognition and positioning,and further determine the work requirements of the manipulator through the sugarcane bud recognition scheme,and design the three-dimensional manipulator The model,and its kinematic analysis and simulation,provide a theoretical basis for the research and development of intelligent planting machinery that meets the agronomic requirements of sugarcane planting.The main research contents and conclusions are as follows:(1)Design a sugarcane seed orientation adjustment system and collect sugarcane seed images.Starting from the sugarcane seed orientation recognition system that combines machine vision and mechanical structure,the essence of the sugarcane bud visual inspection is the positioning of the sugarcane bud centroid;the sugarcane seed has a different initial posture when it enters the visual recognition area.Different sugarcane seed attitude adjustment methods are formulated;based on this,the evaluation index for the success of sugarcane seed recognition and the image acquisition method of sugarcane seed are formulated;the characteristic parameters of the sugarcane object are measured.(2)The identification and positioning of sugarcane buds.In response to the requirements for the position of the sugarcane buds when adjusting the direction of the sugarcane seeds,two different methods for the recognition and positioning of the cane buds are proposed:traditional image recognition methods and convolutional neural network image recognition methods to analyze the characteristics of the sugarcane seed pictures in different color space models,Determine the typical characteristics of the S component of the sugarcane seed in HSV to identify and locate the sugarcane bud;improve a classic convolutional neural network structure,identify and locate the cane bud based on the convolutional neural network,and adjust the better parameters.(3)The design and motion simulation of the sugarcane seed orientation adjustment manipulator.Combining the sugarcane bud visual recognition scheme to determine the work requirements of the sugarcane seed orientation adjustment manipulator,design the three-dimensional model of the manipulator,carry out the positive kinematics analysis calculation and simulation of the manipulator structure,and verify the feasibility of the sugarcane seed orientation adjustment manipulator.The two sugarcane bud identification and location methods proposed in the research were carried out in the sugarcane bud identification and location experiment.The success rate of the identification and location method based on the S component in the HSV model was90%,and the average time for identification and location of each image was 5.58s;The success rate of the convolutional neural network method for sugarcane bud recognition and positioning is 92%,and the average time for each image recognition and positioning is 1.2s,which is better than the convolutional neural network method.The positive kinematics analysis and simulation of the designed sugarcane seed orientation adjustment manipulator verify the feasibility of the manipulator.