Research On Wheel Hub Sorting Technology Based On Machine Vision

With the increasing demand for automation of wheel hub identification and sorting in factories,the use of manual identification and sorting of wheel hub models has problems such as high cost and low efficiency.In addition,the wheel hubs are heavy,and the sorting of them puts a heavy load on the workers,and errors will inevitably occur.Therefore,in order to improve production line production efficiency,increase recognition accuracy,and reduce costs,a set of automatic wheel hub identification and sorting systems with high accuracy,high efficiency,and simple structure are required to meet the automation needs of wheel hub sorting.Machine vision is used more and more widely in industrial production,and it has the characteristics of high recognition accuracy,high efficiency,and non-contact.Therefore,this paper uses machine vision technology to design a set of automatic wheel hub model identification and sorting system,through the research and application of image acquisition,preprocessing,extraction of feature information,classification and recognition technology,the system realizes the classification and recognition of the wheel hub,in addition,PLC is used as a controller to design a sorting system for wheel hub.The main work contents of this paper are as follows:1)According to the characteristics of the wheel hub to be tested,the accuracy requirements of the inspection and the working characteristics of machine vision,the camera and lens and other hardware were selected and an experimental platform was built.In order to ensure the accuracy of the extracted parameters,the camera is calibrated to correct the distorted image.Designed a sorting platform that sorts the wheel hub after identifying the wheel hub model.According to the process of the wheel hub sorting system,a human-computer interaction page and back-end program are developed to monitor the detection results and the status of the system in real time.2)Research on image preprocessing algorithms.Image denoising,image enhancement and target extraction algorithms were studied and compared to select the algorithm suitable for wheel hub image preprocessing.When the filtering method of image denoising is determined,the filtering method is compared and tested,and the improved median filtering method is determined.In the research of image enhancement,the relevant parameters are determined by analyzing and comparing the gray histogram of the background image and the wheel hub image,and the image is enhanced by the method of contrast stretching.Select the minimum cross entropy method to segment the image and then use morphological processing to eliminate the defects in the segmented image.3)According to the shape of the wheel hub,it is decided to extract the diameter and height of the wheel hub,the number of spokes,the area ratio of the holes,the window and the roundness of the wheel hub as the characteristic parameters,and analyze and research the algorithms used in extraction.The k-nearest neighbor algorithm used in classification and recognition has been studied and improved.4)Conduct experiments and analyze the data.Collect and classify the image of the wheel hub,compare the experimental results of the classification and recognition algorithm k nearest neighbor algorithm when the k value is selected differently,when the k value is 5,the recognition accuracy rate is 98.57%.And the sorting test result of the wheel hub is consistent with the recognition result.