Research On Quality Testing Technology Of A Utomotive Gluing

In the production of automobile manufacturing industry,the defects of automobile glue coating are enormous,even can do great harm to human life.The defects of automobile glue coating refer to the defects such as leaking,breaking,too wide and too narrow in the process of glue coating caused by technical factors in actual operation.This paper studied the detection technology of automobile glue coating based on machine vision,aiming at providing an efficient and practical detection method of automobile glue coating for the automobile manufacturing industry,as well as helping enterprises to better control the quality of glue coating,along with save cost and increase productivity.Based on the automatic identification system of defects in a certain automobile manufacturing company,this paper studied a set of algorithms which can classify the defects of automobile glue coating on the basis of perfecting the system function.In the process of automobile glue coating detection,the hardware equipment of the system was selected firstly.Then,the image processing,feature extraction and analysis algorithms were studied according to the characteristics of automobile glue coating on the front cover.The quality detection software of automobile glue coating was developed in line accordance with the functional requirements.Finally,the performance verification experiment of the vision detection system was carried out.Aimed at the defects such as leaking,breaking,too wide and too narrow in the process of glue coating,the following steps were taken.Firstly,the gluing image was collected by the coating machine hand.Then the image preprocessing algorithm of glue coating quality detection technology was studied.After comparing and analyzing common processing methods,the most suitable preprocessing method was determined.Different feature extraction algorithms were analyzed,and the algorithm of feature extraction was determined by analyzing the experimental results.The principle and basic algorithm of support vector machine(SVM)classifier were used to identify and detect the glue coating of vehicles,and the kernel function type with the highest recognition efficiency was determined by experimental data.The experimental results showed that the designed acquisition device can meet the experimental requirements of this paper,and the kernel function and feature extraction algorithm which suitable for this paper were selected for the recognition of SVM.The detection effect was very good and the expected recognition rate was achieved.The real time and detection rate of automobile glue coating in production can be improved greatly.