基于Halcon深度学习的FPC铆点缺陷检测OA

Manual inspection of flexible printed circuit(FPC)rivet defects suffers from low efficiency,while traditional machine vision methods,which rely on features such as rivet area and width for defect classification,often result in high misclassification rates.To address these issues,a FPC rivet point defect detection method based on Halcon deep learning was proposed.First,Halcon operators were used to perform grayscale transformation,threshold segmentation,morphological processing,feature extraction,center point localization,and ROI region segmentation on the defect image to obtain a defect dataset.A lightweight MobileNetV2 network model constructed with Halcon operators was then used to train the dataset to obtain a trained model.Finally,the trained model was imported into the detection software to inspect images captured by industrial cameras and display the final judgment results.This method effectively solves the problem of high misjudgment rate in traditional machine vision methods due to the use of rivet area,width,and other features for defect determination.In the comparative analysis of detection effects with AlexNet and ResNet-50 models,the MobileNetV2 model not only shows a lower error rate but also takes less time when detecting the same ROI area.In actual detection,the accuracy rate of FPC rivet point detection reached 97.40%,and the detection time for a single image is less than 1 second,meeting the task requirements of FPC rivet point defect detection.