Development Of Punching And Bending Quality Detection System Based On Binocular Vision

With the development and progress of industrial technology,the fourth industrial revolution has entered the era of Cyber-Physical Systems(CPS),and the improvement of automation of stamping production lines should also be accompanied by the time to accelerate the pace of reforms.Whether the automatic detection of stamping parts can be achieved is an important indicator to judge the degree of automation.Du e to the low efficiency,high cost,poor stability,and inability to discriminate high-speed objects,traditional manual detection has become a shortcoming in controlling the automated development of stamping production lines.The rapid development of mach ine vision technology in recent years has been widely used in product detection due to its advantages of quick access to large amounts of information,automatic processing,and easy integration of the same design information,processing control information and so on.Therefore,the application of machine vision to the automatic detection of stamping production lines has certain engineering significance and practical value.In this paper,a stepped stamping part is taken as the object of detection,and the relevant theories and techniques for the detection of punching and bending quality problems are studies in depth.The image processing algorithms based on Halcon are discussed in detail.A punching and bending quality detection system based on binocular vision is developed,which is different from monocular vision can only be limited to a certain fixed plane.Binocular vision has richer visual information than monocular vision.A detection method is proposed that uses orthogonal binocular vision to detect t wo surfaces simultaneously,and multi-directional detection can improve the quality detection efficiency.The main research content of this paper is as follows:1.Firstly,through analyzing the quality problems caused by the punching and bending process of the stepped stamping parts,the size of the holes and the bending angle as the detection target objects are determined.According to the principle of machine vision,the detection system hardware structure is given,the selection of light source,indust rial camera and industrial lens are completed,and the binocular vision detection system hardware platform is set up.2.Secondly,in order to achieve the extraction of detection target objects,this paper proposes a method for target extraction of the stepped stamping part and discusses the image processing algorithms involved in detail.The top view of the punching plane is collected by No.1 camera and the side view of the bending plane is collected by No.2 camera.The Blob analysis tools are used to separate the foreground and the background,extract the area where the stamping part is located in the two images.To make further use of the morphological processing,collect the boundary area of the part.The Canny filter is used to extract the sub-pixel edge profiles in the boundary area.Segmentation of lines and circles is through the Ramer algorithm and fitting is through the least squares.This process achieves accurate extraction of the position of the holes and the line segments on both sides of the bending angle.3.Next,in order to improve the accurate measurement,this paper discusses a single camera calibration based on Halcon.Through the analysis of the camera perspective model,the camera calibration is to determine the camera internal and ext ernal parameters.Using the camera calibration assistant,the correspondence between the image coordinates and the world coordinates is established,and the image correction is performed to achieve the measurement of the sub-pixel accuracy and the size of metric units.4.Furthermore,in order to achieve the interactive use of quality detection system and customer,this paper has shown by combining t Halcon and VC++,the system software of the punching and bending quality detection is developed.The detection flow chart and interface design diagram of the software are given.In the software to achieve three major function of camera acquisition control,PC and PLC communications and off-line detection.Finally,the testing results show that the system can see the video collected by the two cameras in real time,control the output signal steadily to the PLC,and can be compatible with Halcon-derived programs to measure punching size and bending angle.The results are displayed in the image display window,which provides intuitive detection and measurement results,providing the basis for the quality detec tion.The innovative point of this paper is to adopt two cameras with orthogonal distribution,as the same time,collect the stepped stamping parts at the same time in dual view.For the quality problems in the production process of punching and bending,a solution based on binocular vision to detect hole dimensional accuracy and bending springback is proposed.