Design And Implementation Of Printing Label Quality Defect Detection System

As the mark of goods,printed label is of great significance to the display of its appearance and content.Environmental interference leads to defects in the label produced and printed.The traditional human eye observation method of removing defect labels is inefficient and unstable,so using digital image processing to detect the defects of printed labels is of great significance.Based on this,designed and implemented a quality defect detection system for color printing labels is.It mainly includes:(1)Firstly,this paper investigated the existing circumstances of printing label defect detection,and analyzed the difficulties of its realized.Then introduced the related hardware equipment and software implementation scheme.(2)Complete the noise reduction,segmentation and registration of the printed label image.On the aspect of noise reduction,the practical effect of several common noise reduction algorithms is tested and analyzed;in the aspect of segmentation,the automatic whole segmentation of label image is realized,and to reduce the errors,the label content is further segmented,and the label content segmentation based on GMM and HSV color space;On the aspect of registration,implemented and improved two registration algorithms.(3)Complete the defect detection of printed label image,including common defect and color defect detection.For common defect detection,implementation of two defect detection algorithms,including image difference method and similarity detection algorithm.The elimination of the false image of differential printing label image is completed and several similarity detection algorithms are tested,some similarity comparison algorithms are analyzed,including hash similarity algorithm,gradient similarity,and structural similarity.For color defect detection,implemented two methods,including based on CIE2000 color difference formula and histogram cross kernel.(4)Finshed the print label defect classification.Firstly,divided the common defects into three kinds,and then collected characteristic parameters of defects.Finally,the label defect classification based on euclidean distance is realized for the collected data.(5)Finally,testing printing label quality defect detection system,including defect recognition rate and speed.Combined with test results,give some performance optimization suggestions.