Research On Mobile Phone Screen Defect Detection System Based On Image Processing

As an important part of the mobile phone,the screen has become the main human-machine interface.However,due to many uncertainties in the production process,there will inevitably be defective screens produced,so the screen quality inspection is indispensable.The current main methods of mobile phone screen detection are manual detection and traditional machine vision detection.The manual inspection method is subjective and easy to fatigue.Meanwhile,it can't guarantee the efficiency and accuracy.Traditional machine vision inspection methods suffer from poor feature extraction capability,low compatibility and high time consumption.To address these problems,this paper proposes a mobile phone screen defect detection system using Siamese convolutional network as the core of the algorithm,and constructs a relevant database system for information management and model updating.This paper turns the defect detection task into a multi-classification problem,in which the screen is classified as free of defects and containing specific defects.In this paper,the defect classification and feature analysis are performed for the collected screen defect dataset.This paper proposes a deep learning algorithm based on Siamese network to solve the problem of easy confusion between different classes of mobile phone screen defects and scattered features of the same class.By incorporating the similarity measurement into the classification task,the requirement of the intra-class compactness and inter-class separability is met.The contrastive loss is added to the back propagation of model training and improved according to the actual situation to make the model training more scientific.Based on the new network model,this paper designs and implements a mobile phone screen defect detection system that combines model training and detection,and improves the data set expansion and image pre-processing techniques.The experimental results show that the proposed mobile phone screen defect detection system has good detection effect and has great application value and development prospect.Based on the need of defect detection research,this paper establishes a mobile phone defect database using relevant computer technology.The database includes the functions of image data management and user information management,and realizes the upload,query and modification of mobile phone screen data.The system is also able to update the Siamese convolutional network proposed in this paper using existing data in database.By testing its related functions,the system can effectively help enterprise employees and research staff save time and resources in data management and sharing,and improve the accuracy of the system,which proves its effectiveness and practicality.