Reaserch On The Control Of Agaricus Bisporus Picking Based On Image Recongnition

In the entire production process of Agaricus bisporus,the picking operation is the most time-consuming and labor-intensive part.With the development of science and technology,the introduction of picking robots in the Agaricus bisporus picking process to achieve a mechanized and intelligent picking process can effectively improve the harvest efficiency of Agaricus bisporus.Not only with practical significance,but also an inevitable trend in agricultural development.The working system efficiency of Agaricus bisporus picking robot is directly affected by the accuracy of fruit recognition and positioning.In this paper,after combining the characteristics of the agricultural picking robot,the hardware system of the Agaricus bisporus picking robot is constructed,and the four-degree-of-freedom robotic arm is modeled and solved.In addition,the identification and positioning of the Agaricus bisporus are related.Research on the algorithm can finally accurately identify Agaricus bisporus and locate its spatial position,providing important information for the autonomous operation of the picking robot.The main research contents of this paper are summarized as follows:Firstly introduce the research significance of agricultural picking robots and the research status at home and abroad,and combine the growth environment of Agaricus bisporus and the operating characteristics of the robot to build the hardware system of the picking robot.The paper uses the improved D-H algorithm to model the four-degree-of-freedom robotic arm,uses the coordinate transformation to establish the positive kinematics equation,and then uses the algebraic method to find the inverse solution.The MATLAB software is used to simulate the forward and inverse solutions.The Monte Carlo method analyzes and simulates the working space of a four-degree-of-freedom manipulator,and finally uses a fifth-degree polynomial interpolation method to plan the joint space trajectory.For the identification of Agaricus bisporus,it is necessary to first perform filtering pre-operations on the image to remove noise in the image,and then use K-means clustering algorithm to segment the target fruit and background,and then further process the image through mathematical morphology.For the identification of the target fruits of overlappingAgaricus bisporus,first the original image is filtered,then the method of extracting the target fruits is analyzed,then the position of the single target fruit is obtained by the K-means clustering algorithm,and the accurate boundary of the target fruits is finally obtained by the improved watershed algorithm information.Recognition of mature Agaricus bisporus targets The normalized Cross-correlation function(NCC)is used to identify the mature Agaricus bisporus,and then the Hoff circle transform is used to obtain the mature fruit position of the Agaricus bisporus.Combined to realize the identification of mature fruits of Agaricus bisporus.According to the principle of binocular stereo vision,Zhang Zhengyou’s calibration method is used to calibrate the internal and external parameters of the binocular camera.This paper presents a centroid-based feature point matching algorithm,and performs stereo matching on the target fruit image pairs of Agaricus bisporus according to the matching constraints.According to the camera calibration parameters,a parallax-based method is used to obtain the target fruit depth value,and the three-dimensional spatial positioning of the target fruit of Agaricus bisporus is realized.Through multiple picking experiments on the target fruit of Agaricus bisporus,the experimental results show that the end effector locates the target fruit of Agaricus bisporus accurately and the picking trajectory is stable,thereby verifying the feasibility of the autonomous picking operation method and the inverse solution algorithm and identification and positioning algorithm Correctness.