Research On Workpiece Recognition And Positioning System Based On Machine Visio

With the further deepening of China’s industrial process,the batch size of is increasing,and the process requirements in product production are also becoming more and more refined.In traditional manufacturing enterprises,the relevant testing in product production is often carried out manually or by semi-automated mechanical operations.The manual operation has large errors,high work intensity,and high costs,while the use of semi-automated mechanical operation can only be performed singularly,with poor flexibility,and cannot meet the requirements of fast,efficient,and intelligent automated production.In this thesis,from the specific application of industrial production,a set of product identification and positioning detection system based on machine vision is designed,around which the following work is mainly accomplished:(1)Overall calibration of machine vision systems.The camera vision model is established,the camera is calibrated by Matlab,the internal and external parameters of the camera are obtained and its projection error is obtained.According to the actual situation,the eye-to-hand model is used between the camera and the robot,and the hand-eye calibration method of the model is optimized to quickly obtain the transformation matrix between the robot coordinates and the camera coordinates.The parameters obtained through camera calibration and hand-eye calibration lay the foundation for subsequent positioning.(2)Image preprocessing and contour extraction.First of all,a reasonable image preprocessing scheme is designed in combination with the characteristics of the detected workpiece,and the effects of the three filters are analyzed and measured by the SSIM and PSNR indicators,and finally,the bilateral filter with the better effect is selected.The effect of the commonly used segmentation method is compared and analyzed,and the experiment shows that the effect of the OTSU segmentation method is better.Then,aiming at the low efficiency and poor resistance of the OTSU segmentation algorithm,this paper proposes an improved OTSU image segmentation algorithm.The difference image before and after filtering is used as the input of the OTSU algorithm,and the improved Seagull algorithm in this paper is introduced into the OTSU algorithm.Through the experimental results,the improved OTSU algorithm can extract the trademark outline information of the research content in this paper.Finally,the segmented image is processed morphologically to remove redundant isolated points and eliminate glitches and holes.(3)Aiming at the difficulty of selecting an appropriate threshold for accurate matching recognition under complex working conditions in Hu moment matching,and the problem of feature point mismatch and time-consuming problems in SIFT feature matching under complex working conditions,this paper proposes a method based on global Hu matching and local SIFT matching.The fused Hu-SIFT fast matching method,and aiming at the problems of the fused Hu-SIFT algorithm,an improved Hu-SIFT algorithm is proposed to improve the matching accuracy,reduce the calculation time,and meet the needs of real-time detection.For the positioning requirements in workpiece detection,the spindle rotation method is proposed to reduce the time complexity of the minimum circumscribed rectangle,and the positioning method based on the combination of the auxiliary template and the minimum circumscribed rectangle is used.(4)A product inspection and positioning system is designed.The main functions of the software were specifically analyzed,and a product vision inspection software system was prepared based on the VS2019 development environment,the MFC Windows platform application development framework,and the Open CV open-source graphics processing library.The design,installation,and commissioning of the hardware were completed.The vision system calculates the positioning data for transmission to the 4-axis robot for precise positioning.The system achieves intelligent product online inspection and process positioning,improving the automation level of industrial production of collaborating manufacturers.The system has been used in the production line of Bozhou Fuda Printing Co.,Ltd,and has achieved very good results and economic benefits.