Research On Defect Detection Technology Of SMT Based On Machine Vision

With the rapid development of electronic manufacturing,the connecting between circuit board and components has been widely used surface mount technology(SMT).As the first working procedure of surface mount technology,solder paste printing is a critical step to ensure the quality of PCB products.It will directly affects the subsequent patch,solid coupling and cleaning process.The success of the solder paste printing relies on the SMT steel relief hole correctly or not.Nowadays,the size of SMT steel's hole is becoming more and more smaller.So the traditional manual visual inspection cannot satisfy the requirement of steel net defect detection.Therefore,this paper proposes using computer vision to automatically detect the defect of steel net information.This article will revolve around computer vision detection technology of error compensation,primitive clustering,path planning,image filter processing,subpixel edge detection and so on.This paper carried out a research into the following several aspects:Firstly,this paper puts forward the research background and significance of this topic,summarizes the foreign experiences and lessons in stencil defect detection technology,and analyzes the insufficient in defect detection technology.Based on this,determine the goal of this topic and design a stencil defect detection system which has obvious advantages in detection accuracy and detection speed.Secondly,finished design and processing the mechanical workbench of SMT stencil defect detection system.The mechanical workbench is the system's hardware,so it's accuracy requirement is very high.Based on analysis the reason of mechanical workbench's error,using a laser tracker has carried on the data collection to the workbench,this paper proposes a using quadratic spline difference on mechanical workbench error compensation method.Thirdly,researched the Gerber data in detail,and obtained high precision standard image.Then based on the standard image which obtained from Gerber data,using the improved genetic algorithm to clustering figure image and path planning method.The algorithm is verified by the actual image acquisition experiments on the detection speed advantage.Then,proposed an improved median filter to get rid of the noise pollution.At the same time,proposed an improved Canny-Zernike moment subpixel edge detection method,and realized the sub-pixel's level edge detection.Lastly,completed the SMT stencil defect detection system's design based on the LabVIEW and comprehensive utilization of the programming language such as C/C,MATLAB and so on.Based on the actual image possible rotation,tilt,offset,such as error correction,achieve the final defect detection function of SMT stencil.Experimental results show that the SMT stencil defect detection system in terms of detection accuracy and speed has reached the requirement of practical use,and laid the groundwork for other complex defect detection technology.