On-line Detection Of PCB Components Based On Machine Vision

In recent years,the electronic manufacturing industry has been developing rapidly.Many electronic manufacturing companies introduced the placement machine and plug-in machine in the production of circuit boards,combined with manual operation to place components on the circuit boards.In manual operation work fatigue maybe will cause some operation errors,such as wrong component,missing component and the polarity of the electric component etc..At present,the automatic optical devices for online detection of circuit boards are based on solder joint quality detection,and the detection accuracy of directly inserted components is low,so it is easy to misreport.In this thesis,machine vision based online detection system for PCB components is studied.First,PCB images are collected online,then PCB diagrams are processed and analyzed,and finally the detection results are displayed.The system consists of the image online acquisition module,the image splicing module,the component detection module,the software interface module and the data storage module.The following is the main work of the thesis:(1)Online image acquisition and splicing: For a feature area set up online PCB video image acquisition,image frames and every interception feature area.In this feature area,the PCB image is uniform in light,and the shadow block area of the components is small,which can be used as a splice.The splicing object is output to the image splicing module,and a test map with uniform brightness and small height difference is obtained through the image stitching algorithm.The algorithm use a similar region search algorithm and a mosaic method which based on gray level template matching error correction.(2)Component detection module: this part is made up of template making and defect detection.In template making,PCB assembly files are imported to generate ROI(region of interest,area of interest),and then debug some ROI's special components by manual mode.The circuit board defect detection includes detecting the presence of components and the insertion of polar components.Component Existence Detection matching two algorithms to check error though color image segmentation and color based targets.Component Direction Detection check the detection of diodes,electrolytic capacitors and sockets through diode detect method which based on improved image enhancement method,electrolytic capacitor detect method which based on the circle fitting and the thresholding,socket detect method which based ondimensional metrology.(3)The program interface module: according to the function is divided into 6 interface,its features include user login,defect detection,result display,online splicing,template management and data storage;the use of LabVIEW programming provide a different display interface for different users to avoid irrelevant personnel operate and modify the program.(4)Data storage module: it can save and read the data of template data and detection results.When testing different types of circuit boards,the corresponding template can be built or loaded.In the detection process,the results are stored in the temporary data table.After the test,the export report can be saved in the local folder.(5)Finally,the functional modules of this thesis are tested,and the proposed method is verified.After online image collection and combination,the accuracy rate of the element detection is up to 99%,and the average detection time is between 1 and 2 seconds.