Research On Image Detection Method For Text Defects In Printed Matter Based On Corner And Block Template Matching

Printing products such as publications,books,labels,packaging of commodity are everywhere and actively promoting the spread of daily culture.However,due to the complexity of printing technology and rapid improvement of printing speed,various defects of surface quality will appear,which is inevitable.If the defects in unqualified printing products are not eliminated,serious consequences may be caused.Therefore,scientific,rapid and effective detection of surface defects of printed matter has become an inevitable trend of industrial development.Manual detection methods often have low efficiency,high labor intensity,poor repeatability and other problems,so it has become the main trend of technology development in the printing industry to use machine vision means to carry out surface defect detection of printed matter on the high-speed printing production line.Based on the above background,aiming at the shortcomings of traditional manual detection and the low accuracy of local matching in visual detection,this paper is oriented to the actual demand of sampling detection in production,combined with the analysis of the equipment vibration,paper flexibility and other interference factors on visual detection,classifies the text defects,and studies the detection technology of printed text defects.This paper presents a method for the detection of printed text defects.In this method,the template image is taken as the detection standard,and the regular line in printed matter is detected by Hough straight line to realize the overall correction of the size and pose of the scanned image.In order to solve the problem of low local matching accuracy,the Harris corner and block template matching method were innovatively adopted.On the basis of determining the block size by the block coefficient experiment method,the matching feature points were further expanded,and the high-precision local matching between the template image and the scanning image was realized,and the local correction of the scanning image was completed.Then,the positive and negative difference and fusion technology are used to obtain the image of missing print,ink spots and ink imbalance.Finally,through the connected domain analysis of Blob,pseudo-defect removal,similarity measurement,classification of defects and other image processing technology,the printed text defect detection is realized.In order to facilitate user’s development and promote the process of print defect detection,this paper briefly introduces the development environment of the detection system,and designs the application interface.The experimental results show that the detection rate of printed matter defects is94.1% and the false detection rate is 3.0%.The detection rate of intermediate precision class was 91.3%,and the error rate was 18.4%.The detection rate of low precision grade was 88.2% and the error rate was 37.6%.For A4 paper,the average detection time of this method is about 5.6 seconds.This method can well meet the requirements of production sampling,and the research results have certain application value in the printing industry.