Research Of Defect Detection Technology In Printed Sheets Of Automobile Dashboards

In view of the traditional human eye detection in the printed sheets of automotive dashboards with problems of the low efficiency,high labor costs,and subjecting to man-made factors to an easy false detection and missing detection,it can not meet the needs of large-scale production.Analyzed the defects characteristics of the printed sheets of automotive dashboards fully and combined the actual production and application scenarios,this paper designs and implements a set of defect detection system about automotive dashboard.The main research results are as follows:(1)A fast and high-precision automotive dashboard image registration algorithm is proposed.In the algorithm,the background of image redundancy is removed firstly and the geometric features of the foreground target contour are extracted.Next,the rough matching features are picked by measuring the cosine distance.And then the precise matching features are obtained by the RANSAC algorithm to eliminate the false matching.At last,the registration parameters are computed to complete the registration.The experimental results show that for the dashboard image of 3840×2748 pixels,the registration algorithm proposed in this paper has a low sensitivity to illumination changes,a high registration accuracy and a short registration time.In addition,the longest registration time does not exceed 2 seconds and the registration deviation is within 1 pixel.(2)The edge detection algorithm is combined with the template difference image method to effectively extract the defect information.Simultaneously,the SVM algorithm is used to classify the five kinds of dashboard printing defects.Finally,the detected defect information stored in the database can be regularly managed which can provide a great reference value for the later improved production line.(3)A dashboard defect detection system is designed and implemented.The detection speed of the system can be achieved within 20s/disc.the correct rate of defect extraction be achieved 90% and the correct rate of defect classification be achieved 90%,which can meet the actual industrial inspection requirements.The system has been put into the enterprise application testing.