Application Research On Defect Detection And 3D Reconstruction Of Gas Cylinder Inner Wall Based On Binocular Stereo Vision

With the development of social industry level,the application of metal gas cylinders in daily life has become more and more extensive and various types,which mainly include double-end welding gas cylinders,high-pressure metal gas cylinders,low-temperature storage gas cylinders and other products,The relevant detection of its safety production and product quality is becoming more and more important.In recent years,the application and development of machine vision imaging technology in the field of China's industrial product quality defect detection technology has continued to mature,its work efficiency is high,and its reliability is strong.The advantage of moderate cost is gradually reflected.Therefore,for the detection of the inner wall of metal gas cylinders,this article designed a defect detection and three-dimensional reconstruction system for the inner wall of the gas cylinder based on binocular stereo vision.The design system mainly solves two major problems:First,detect the defects of the inner wall of the gas cylinder and perform qualitative analysis With quantitative calculations,second,the three-dimensional shape of the inner wall of the gas cylinder is reduced by binocular stereo vision technology.The main research work is as follows:(1)Design the entire gas cylinder detection system plan,determine the platform device selection and software functions of the detection system.Adopt mobile scanning imaging to realize the defect detection of the inner wall of the entire gas cylinder.According to the defect detection requirements,determine the relevant important performance parameters of the binocular camera,determine the camera sensor type is CMOS,and the data transmission between the camera and the computer is a USB serial port;use a point laser and a flat cone to assemble a laser ring projector,The function is to project the laser on the inner wall of the gas cylinder to form a laser halo to assist detection;determine the type selection and clamping support structure of the electric translation stage;realize the construction of the entire hardware platform.(2)In-depth study of the imaging principle of the binocular stereo camera,and completed the calibration of experimental camera parameters in the system.Introduced the detailed principle of Zhang Zhengyou's 2D checkerboard calibration method,analyzed the meaning of internal and external parameters and distortion parameters in camera calibration;selected MATLAB software tools for the dual target calibration experiment of this subject,analyzed the calibration average error,and determined the vision system camera Key performance parameters.(3)Design a defect detection algorithm based on the geometric features of the laser ring.First,pre-process the inner wall image of the gas cylinder through the ACE image enhancement algorithm and grayscale;then use the Steger halo center extraction algorithm to complete the sub-pixel positioning of the center of the light bar,and then perform image denoising;Halo characteristics,targeted design detection algorithm;according to the defect category,design multiple sets of comparative experiments,use the defect algorithm of this article to process separately,analyze the detection results,in order to initially verify the feasibility and accuracy of the defect algorithm.(4)Three-dimensional reconstruction of the inner wall of the gas cylinder and analysis of the experimental results of the visual inspection system.First,the binocular polar line correction is performed on the binary image of the extracted laser light halo;the parallax value is obtained according to the corresponding matching relationship between the points on the left and right halo;according to the binocular stereo principle,the three-dimensional point cloud data corresponding to multiple sets of halos on the inner wall are calculated The three-dimensional reconstruction of the inner wall of the gas cylinder is completed by the greedy projection triangulation algorithm;combined with the detection results of the algorithm,the defect detection algorithm designed in this paper has a good recognition rate for single defects on the inner wall of the gas cylinder,and also has a clear distinction between the presence and absence of the inner wall image The comparison between the calculated value of the defect parameter algorithm and the actual measured value is carried out in multiple groups.The absolute error value of each group is within 1mm,which meets the expected defect detection accuracy requirements and verifies the stability and measurement accuracy of the software algorithm in the design scheme.