Research On Online Recognition System Of Wheel Hub Type Based On Machine Vision

With the popularity of automobiles,the demand for automobile wheel hub has also increased.Wheel hubs are also becoming more diverse.In the mixed flow production of multi-varieties and small batches of wheel hub,the traditional way is manual recognition,which has the shortcomings of artificial fatigue and low recognition efficiency,and at the same time,the demand for production automation is constantly increasing.Therefore,the application of machine vision technology in the automatic production of wheel hub and the replacement of manual recognition by machine are of great significance to the reduction of production costs and the improvement of production efficiency.Aiming at the problems of traditional manual recognition and image recognition,this paper studies the method of online recognition of wheel hub blank type by three-dimensional point cloud processing technology.The specific contents are as follows:Firstly,according to the configuration characteristics and conveying forms of wheel hub blank,the advantages of three-dimensional point cloud recognition over image recognition are compared,and the system scheme and composition for online recognition of wheel hub type are put forward,and the working flow of the system is analyzed.Then,aiming at the problem of large amount of wheel hub point cloud data and the existence of a varity of noise points,we analyzed the causes and characteristics of noise points.After that,the downsampling algorithm and filtering algorithm of wheel hub point cloud are studied.An adaptive segmentation algorithm based on depth feature is proposed to remove burrs from wheel hub point clouds.The feature extraction algorithms of wheel hub such as diameter,width,rotation period,number of window and spoke shape are studied.The inclination and motion fluctuation of wheel hub are judged during the feature extraction process to determine whether the wheel hub point cloud data is available.According to the spoke rotation period and the number of window,the wheel hub is divided into different search spaces.And then,through SVM classification algorithm and shape context matching algorithm,the automatic recognition of wheel hub type is realized.Finally,an online recognition system for wheel hub type is developed by using MFC library based on C++ language,PCL library based on point cloud and OpenCV library.An experimental platform is built with hardware such as line laser sensor,photoelectric sensor and single chip microcomputer.Finally,the accuracy of feature extraction,recognition accuracy and recognition efficiency of the system are tested.The experimental results meet the recognition requirements and verify the feasibility of the methods proposed in this paper.