Defect Detection In Product Assembly Based Machine Vision

With the global attention to the intelligent manufacturing and the development of machine vision in recent years,the introduction of machine vision testing into the industrial production process in order to achieve more accurate and efficient detection of the purpose of national industrial development has become an important task.In this paper,the assembly defect detection,as an important part of the electronic manufacturing production line,aims to use the machine vision method to solve the product production process of assembly quality problems,to control the production quality and reduce production costs.In this paper,the assembly defect detection is mainly used in the assembly of 3C electronic products.Barcode assembly defect detection mainly check whether the product or the circuit board has been put on barcode as expected.The barcode information will be uploaded to the production record when it's integral,or the alarm will be prompted to be handled by the operator.The terminal assembly defect detection is to check whether the terminal plug in place,whether there is poor contact.Barcode assembly defects in this work focus on the complex background and the lighting conditions are not ideal.The barcode segmentation mainly uses the regularity gradient feature of barcode to design a segmentation method based on gradient histogram statistics.The method can segment the barcode very accurately,and the efficiency is also high,and the multi-barcode extraction is supported,but the method exists when the barcode and background discrimination is not high,the barcode background compared to the whole image serious dark when the poor results of the situation.In order to solve this problem,the paper designs a bottom-up barcode segmentation method based on local gradient information fusion,which avoids the failure of segmentation failure due to the effect of foreground extraction.The barcode reading is based on the dynamic threshold of barcode bar space information definition method to solve the barcode illumination uneven problem.Terminals assembly defect detection is divided into positioning and measurement of two parts.This paper designs two kinds of terminal positioning methods,one is to match the approximate position of the template and then through the feature point to determine the exact direction of the method,the other is the use of pyramid ideas from coarse to fine rotation template matching positioning terminal method.In the measurement part,the Hough straight line fitting method based on the gradient extreme edge is designed to determine the defect.Compared with the traditional Hough straight line fitting,this paper changes some of the fitting details to suit the current application scene.The experimental results show that the design of the assembly defect detection method is very effective for the barcode and terminal detection under the current application,and has high detection rate and low detection rate.The running time also satisfies the existing application claim.