Research On Wrorkpiece Recognition And Manipulator Positioning And Grasping System Based On Binocular Vision

In the modern industrial production line,the automatic grasping of workpieces is a very important part of intelligent production.The binocular vision technology can use the parallax principle to recover the spatial information of the target object and combine the manipulator with it.This article is based on the needs of robots in the field of intelligent manufacturing to achieve accurate and efficient automatic gripping of workpieces.Based on binocular vision of workpiece recognition and robotic positioning and gripping system research,the robotic gripping process is optimized and improved.First of all,this article studies the manipulator grasping system,establishes the framework of the grasping system,selects the parallel binocular vision model,introduces the software required in the grasping system and the hardware for building the system,and describes the article based on binocular vision.Relevant technology in the workpiece picking system.Secondly,the factors affecting the image quality are analyzed,the types of noise are discussed,the wavelet method for removing image noise is studied,the wavelet mathematical model for image noise removal is created,and the workpiece image adaptive threshold denoising algorithm based on wavelet is used.In the wavelet transform,the value of the coefficient threshold changes with the change of the wavelet decomposition layer,and the coefficient selection method adjusts itself according to the change of noise.Through experimental observation,it is found that the noise image processed by the improved wavelet algorithm proposed in this paper is more thorough and retains more target details than the noise image processed by other denoising algorithms.Add different degrees of noise to the workpiece image in this paper,and conduct a denoising comparison experiment with the traditional denoising algorithm.According to the experimental data,establish a denoising image quality line chart for data analysis to verify the robustness of the improved wavelet transform algorithm.By analyzing the influencing factors of the workpiece image matching process,the SIFT workpiece image matching algorithm based on Canny and wavelet improvement is adopted;the SIFT algorithm feature selection process is studied,and the feature point selection process of the SIFT algorithm is optimized by combining the features of Canny edge extraction.Combined with the characteristics of multi-scale wavelet transform,the image matching process of SIFT algorithm is optimized.The feasibility of the improved SIFT algorithm is verified by experimental data,and the conclusion that the improved SIFT algorithm proposed in this paper is shorter than the traditional algorithm and has higher accuracy.Then,the target is identified and segmented and the centroid of the workpiece is calculated using the centroid calculation method.The camera calibration experiment was carried out using Zhang Zhengyou calibration method,and the internal and external parameters of the camera were obtained.The basic principles of 3D reconstruction are introduced,and the pixel coordinates are 3D reconstructed by combining the calibration results and the parallel binocular vision model.Finally,the D-H model operation calculation was performed on the movement process of the manipulator,the manipulator grasping platform was built,the MFC-based workpiece recognition and robot grasping system interface was developed,and the vision system and the grasping system were combined on the constructed experimental platform to carry out Crawl experiment.By analyzing the grabbing experimental data,it is concluded that the system grabbing success rate constructed in this paper is above 86.7%,and it is verified that the binocular vision-based workpiece recognition and manipulator positioning grab system can meet the target workpiece image recognition and robot real-time grabbing Fetch requirements.