QR Code Recognition Of Industrial Products Based On Machine Vision

In the inspection and sorting system of modern industry,the manual code scanner used to detect the QR Code of industrial products has slow manual recognition speed and low efficiency,and the existing QR code industrial cameras based on machine vision There is a need to further improve the recognition rate of automated recognition systems in complex backgrounds.In this thesis,the HDY-QR detection module is designed according to the amount of data to be tested,and a prototype of a QR code detection system based on machine learning detection algorithm is developed,which improves the QR code recognition accuracy and reduces the detection time,which is helpful for improving industrial automation and intelligence.The automatic detection of QR codes of products in manufacturing provides a certain research basis and application value.First,for the small amount of data to be tested,this thesis proposes a QR detection method based on HOG-GB+SVM.When the amount of data is small,it is suitable for manual feature extraction and the detection accuracy is high.Therefore,starting from image feature extraction,Gaussian filtering and bilinear interpolation are introduced to improve the feature extraction ability of HOG for QR codes,and the SVM classifier is combined to improve the detection accuracy in scenarios with small datasets.Second,when the amount of data is large,the detection methods based on traditional manual feature extraction rely too much on preprocessing,and high-dimensional data cannot determine the kernel function of SVM.Therefore,this thesis designs and implements a deep learning-based QR code detection module.Among them,the Dense Net detection model is used,and the dense connection unit is used to improve the detection speed under the condition of high detection rate.In addition,since the end-to-end structure of the Dense Net model cannot use feature fusion to improve the detection accuracy and cannot meet the detection requirements when the QR code image is noisy and blurred,the YOLOv5s-based detection model is used,and the FPN+PAN module in the model structure is used to To better extract features,embed the SCConv structure into the Backbone area,improve the feature extraction capability of the Backbone area,and at the same time appropriately simplify the Neck area of the YOLOv5 s model,thereby reducing the model complexity and enhancing the detection efficiency of QR codes.According to the requirements of intelligent manufacturing for fast QR code detection and high precision,determine the overall functional requirements and visual hardware of the QR code detection system.The detailed design of each module of the QR code detection process is given,and the selection of visual hardware modules(camera,light source,lens)and the construction method of the machine vision motion platform are determined.This thesis uses Python+QT to develop and implement the prototype of the QR code detection system,and build a physical experimental device that simulates the detection of QR codes in an industrial assembly line.Finally,in order to ensure the stability and availability of the system,combined with the hardware platform,the HDY-QR detection module is tested,and the performance test of the system prototype and the function test of each module are carried out.The test results show that the prototype of the QR code detection system implemented in this thesis runs stably,obtains better detection results in multiple scenarios,and meets the needs of enterprises for the speed and efficiency of QR code detection.