The Research On The Mixed Line And Defect Detection For Automotive Glass Based On Machine Vision

Automotive glass is an integral part of the automobile;its quality directly affects the appearance quality of the car and affects drivers’ driving safety.In the actual production line,automotive glass detection mainly involves two aspects: mixed-line detection and defect detection.The mixed-line detection detects whether the automotive glass transported by the current production line meets the specifications,that is,to see whether the outline,material,thickness of the automotive glass meet the requirements.The defect detection is to detect whether there are quality problems in the automotive glass,mainly including the defect detection of the screen printing part and the defect detection of the automotive glass itself.At present,many glass manufacturers still focus on manual inspection for mixed-line detection and defect detection of the automotive finished glass.However,with the automotive glass industry’s vigorous development,traditional manual testing is gradually unable to meet the growing demand for production capacity.Nevertheless,most of the existing glass detection technologies are for detecting fully transparent float glass,which is not suitable for mixed-line detection and defect detection of automotive finished glass with screen printing.Because of the above situation,this paper analyzes the characteristics of automotive glass and the features of common defects and designs an automotive glass mixed-line detection algorithm combined with contour features and color features and an automotive glass defect detection algorithm based on regional growth,respectively.The main work of this paper is as follows:(1)Image acquisition: This paper analyzes the characteristics of automotive glass itself and determines the information required to be acquired when automotive glass is used for mixed-line detection and defect detection;according to the imaging characteristics of the required information and automotive glass image acquisition device with dual cameras and double background is designed to ensure that the contour information,color information and defect information of the glass can be collected thoroughly and well.(2)Mixed-line detection: A mixed-line detection algorithm for automotive glass is proposed,which combines contour features and color features.Without increasing the difficulty and timeconsuming of the algorithm,a geometric moment matching algorithm combined with area features is proposed to improve contour detection accuracy.(3)Defect detection: This paper compares the advantages and disadvantages of various image segmentation algorithms for automotive glass defect detection applications and proposes the automotive glass defect segmentation technology based on adaptive threshold algorithm and region growth algorithm,which solves the problem that existing segmentation technology is difficult to segment the automotive glass defects intact,and carries on the quantitative detection and classification to the segmented defects.(4)Monitoring software: This paper designs a machine vision-based monitoring software for automotive glass mixing line and defect detection,displaying the whole process of automotive glass inspection in real-time and can complete monitoring,modification,and setting of all parameters in the inspection process.The work done in this paper can effectively realize the mixed-line detection and defect detection of automotive glass.In the laboratory environment,the average success rate of mixedline detection is 98.62%.The average success rate of defect detection reached 97.07%,which achieves the expected design goal and has a certain value of use and promotion.