Research And Realization Of Detection Of Surface Defects Of Daily-use Porcelain And 3D Reconstruction Of Defective Products

As the country with the largest production and demand for daily-use porcelain,improving the quality of daily-use porcelain is a problem that China urgently needs to solve.At present,the surface defect detection of daily-use porcelain is mainly based on traditional manual inspection,which has problems such as low efficiency and high false detection rate,so it cannot guarantee the quality of daily-use porcelain.And only through the surface defect inspection can not fundamentally solve the production quality problems of daily-use porcelain.This subject is based on machine vision to complete the high-speed and automatic defect detection of daily-use porcelain to improve the quality of daily-use porcelain;on the other hand,the three-dimensional reconstruction of the detected daily-use porcelain residues is used to collect the shape of the surface defects,Direction and location information is uploaded to the relevant process designer.In the later stage,process designers can perform data classification analysis,optimize process flow,and adjust production equipment,so as to improve product quality and reduce production costs.The main work carried out in this dissertation is as follows:1.A visual inspection system for surface defects of daily-use porcelain was designed.First,an image acquisition hardware platform for daily-use porcelain was built according to the actual industrial requirements.Then,according to the characteristics of daily-use porcelain production process,a detection algorithm for defects such as deformation,notches,cracks and stains was written,and a software system for detecting surface defects of daily-use porcelain was built using Open CV and MFC platforms.Finally,200 daily porcelain samples were tested experimentally.The test results show that the accuracy of the system can reach 98%.2.A feature matching algorithm combining FAST + SIFT + FLANN algorithm is proposed.In order to obtain the three-dimensional information about the defects of daily-used porcelain,the classic Zhang’s checkerboard calibration method was used to calibrate the binocular camera,and then according to the product characteristics and defect characteristics of daily-use porcelain,this paper proposes a combination of three algorithms for feature matching algorithm.FAST algorithm completes feature point extraction,SIFT algorithm completes direction description,and FLANN implements feature matching.And use RANSAC algorithm to complete iterative optimization,and get accurate feature matching point pairs.Experiments show that the proposed feature matching algorithm can efficiently,quickly and accurately obtain rich matching point pair information.3.A three-dimensional reconstruction system of daily-use porcelain was designed.First,the 3D information of the spatial points is calculated based on the camera calibration results and the matching point pair information.Then use the Delaunay triangulation algorithm and Open GL platform to complete the visualization of the 3D point cloud,and get the 3D shape,direction and position information of the defects of daily porcelain residues.