Research On Detection Method Of Cellphone Screen Glass Defect Based On Machine Vision

Under the trend of increasing popularity of mobile terminal devices,digital products such as notebook computers,cameras,and tablets are becoming more and more frequently used,and mobile phones are indispensable in people's daily lives.The touch screen is an important part of the mobile phone.The quality of the screen glass is directly related to the normal use of the mobile phone.The defects on the glass cover will affect its mechanical and optical properties and seriously damage the use value of the product.Therefore,it is very important to detect defects in the production of glass cover.In view of the fact that the current frequency of mobile phone product replacement is getting higher and higher,and the demand for manufacturing is large,traditional manual testing cannot be satisfied.The application of machine vision instead of the traditional method is also the development trend of industrial production.In view of the above situation,this paper analyzes the glass cover production process and defect detection standards,discusses the common defect types and existing detection theories,designs the overall scheme of the glass cover defect detection system,and compares the different pretreatments.The application of the method on glass defects,in-depth research and discussion on the glass edge defect and internal defect detection classification algorithm,and complete the relevant simulation test.The main research contents of this paper are as follows.(1)According to the production process flow and defect detection standard requirements of mobile glass cover,based on the analysis of defect image data characteristics,a glass defect detection system was constructed.The system consists of a light source imaging system,an image acquisition device,an image processing system,and system software.(2)Contrasting and analyzing the currently used image preprocessing techniques,based on the analysis and analysis of image smoothing filtering,morphological operations and edge detection algorithms,respectively,corresponding to the characteristics of edge defects and internal defects of the cover glass,corresponding to appropriate pretreatment methods,And do simulation experiments on the actual samples,and the results are obtained.(3)For the chipping defects of the glass edge,after pre-processing,the edge data point set is obtained by the method of row and column pixel traversal search,and the simulation experiment results which are consistent with the least squares method and the random sampling are compared,and finally the random sampling consistent method is selected.The edge straight line model is combined,and the edge defect area is determined according to the deviation between the actual data and the fitted model.(4)Based on the characteristics of internal defects of glass,the morphological features and invariant moment features of the defect image are discussed and extracted.In order to get a better training model,the parameter selection and optimization of svm method are discussed,and discussion is made to solve the sample non-uniformity.The method effectively improves the accuracy of the classification.