Machine Vision To Sort Lacerated Magnetic Ring In Artificial Neural Network Algorithm And Its Hardware Implementation

Magnetic surface defect is one of the important factors affecting the quality of the magnetic ring.Traditional manual testing has many disadvantages in magnetic sorting.with the rapid development of machine vision and automation requirements,we need to design a set of magnetic sorting system to conform to the enterprise actual production.This paper research and analyse the common surface defects of lacerated magnetic ring,collect the abundant original image and preprocessed the picture in gray and histogram equalization.After filtering the background and noise,we use threshold binarization and morphological operation to get defect area.The feature space is established by extracting the geometric features and statistical features of the defect part.According to the demand of classifier,the BP artificial neural network model is selected.According to the feature space and the defect type,the three layer neural network is designed,and the processed pictures are grouped into training,testing and programming.The advanced DVS8168 video platform is selected for hardware and the embedded development platform is built according to the internal architecture.The software aspect selects opencv3.2 for image processing on the Visual Studio2015 platform.QT is chosed to complete the user interaction interface.after all,it achieved the Intuitionistic,quick,convenient and efficient magnetic sorting system.