Font Size: a A A

Research On Defect Detection Technology Of Printed Circuit Board Based On Augmented Reality

Posted on:2024-08-12Degree:MasterType:Thesis
Country:ChinaCandidate:B TuoFull Text:PDF
GTID:2568307181451864Subject:Master of Engineering (Electronic Information) (Professional Degree)
Abstract/Summary:PDF Full Text Request
In recent years,with the development of modern industry,electronic products are constantly updated and iterated.PCB(Printed Circuit Board,printed circuit board)is the core component of electronic products,its quality determines the life of electronic products.Due to the complex production process of PCB,the defects of PCB may occur if the production process is not paid attention to,so PCB quality supervision is essential in the process of PCB production.This paper studies the defect detection technology of printed circuit board based on augmented reality,designs an augmented reality system that can accurately identify PCB board and its defects,and superimpose PCB defect information and production information on the PCB board in real scenes to provide users with PCB defects and related information to assist PCB quality management.The main work is as follows:In terms of PCB defect detection algorithm,a new PCB defect detection algorithm based on YOLOX-WSC was proposed in view of the difficulty of PCB defect detection in complex scenes and the problems of false detection and leakage detection.Firstly,in the part of data enhancement,in order to reduce the error of input network data,the method of optimizing data enhancement is adopted to optimize.Secondly,due to the fact that the background of PCB images is mostly similar and the texture is complex,it is difficult to extract features and the loss of feature information in network transmission.Therefore,the effective feature recognition capability is strengthened by introducing parameterless attention Sim AM into the backbone network,using the energy function to identify effective features and suppress irrelevant features.Finally,in order to obtain more higherorder semantic information,improve resolution and increase the fusion and interaction ability of feature fusion network,CSPHB module is used in feature fusion network,so as to enhance the feature fusion of different scales and better improve the spatial interaction of downstream tasks.Through experiments,it can be found that the detection accuracy of the new algorithm is significantly improved.Compared with the current popular algorithms,the new algorithm is superior to other algorithms in accuracy and reasoning speed,which further verifies the effectiveness of the text algorithm.On the method of tracking registration,in order to solve the problem of real-time tracking registration,the paper studies the method of integrating BRISK-GMS and pyramid LK optical flow.In view of the low accuracy and long time spent in traditional feature matching,we use BRISK-GMS for feature detection and matching.We extract and match PCB images with binary feature description operator BRISK and the quick feature description matching method based on grid motion statistics(GMS)to obtain the correct feature point pairs.After successful matching,the point pair was tracked using pyramid LK optical flow,and the camera position and pose were estimated based on the camera internal parameters obtained by the camera internal parameters calibration method,and the registration was completed.Through experimental verification,the method has improved the accuracy of feature matching,and the time consumption also meets the real-time requirements,which can ensure the accuracy while conducting real-time tracking registration.In terms of virtual fusion display,in order to complete the fusion display of virtual model design and real scene,this paper adopts handheld device to collect video images,combines the position information and PCB defect information obtained by tracking registration module,and uses Open GL and other tools to display enhanced effects.Through experimental verification,the system can identify PCB target and superimpose virtual model display,presenting multiple information and smooth interface display,indicating that the system can be used in PCB defect quality management work,with strong practicability.In summary,the designed system can detect PCB defects well by using YOLOXWSC algorithm,and the detection accuracy of m AP@0.5 and m AP@0.5:0.95 increases by2.88% and 11.64%,respectively.Using the tracking registration algorithm that combines BRISK-GMS and pyramid LK optical flow can effectively improve the speed and accuracy of feature detection and complete the rapid real-time tracking registration of PCB.Based on these two algorithms and virtual-real fusion display technology,the fusion of virtual model and real scene is completed,and the enhanced display effect of various PCB information is successfully realized,and the practical application value of the system is verified.
Keywords/Search Tags:Augmented Reality, Image Feature Detection, Tracking Registration, Virtual-Real Fusion, PCB Defect Detection
PDF Full Text Request
Related items