Research And Application Of Printing Defect Detection

Due to the complexity and rapidity of printed process,it's inevitable that printed matter might present multiple defects with appearances.defects in appearance have many kinds,such as under print,missed print and uneven printing ink,defective printed matter are unqualified,it must be culled.Owing to traditional manual inspection is difficult to repeat,and the accuracy level of traditional manual inspection is still low,it's hard to meet demand for high-quality production.Therefore,research of vision defect detection is necessary.This paper designs a printing defect detection algorithm through the study of defect detection technology of printed matter image.this pre-processing applied techniques such as image segmentation,binarization,image correction and registration in the process of processing printed matter images.In defect detection respect,this paper compared several traditional algorithms,which improved the traditional method of image subtraction and realized the detection recognition of defects on image samples to be tested combined with area growing algorithm,and finally accomplished the classification of defect.The main research results are as follows:1.Plan design: this paper designed the plan of printing defect detection method combined with research results at home and abroad and actual industrial needs through the study of current defect detection technology of printed matter image.2.Image preprocessing: printed matter image are segmented to get the region of interest beforehand,and ROI of printed matter image samples are corrected through Hough Transform.Finally,The ORB feature extraction algorithm is used to register the printed matter image samples.3.Defect detection: this paper proposed a detection algorithm based on the traditional method of image subtraction,the algorithm used Average Character Height to quickly generate cells for detection,and perform differential processing on the cell of the samples of template image and the image to be tested through improved method of image subtraction,so as to achieve defect detection combined with area growing algorithm.4.Defect classification: the classification of defect was achieved according to the geometric feature and the grayscale feature of defects that occur frequently ocounted.Based on the research of printing defect detection,a software system for the research of printing defect detection is designed.The software system implements functions such as customized image segmentation,image preprocessing,image defect detection and image defect classification.The experimental verification shows that the recognition rate of defect and the accuracy of defect classification reached more than 97%.