Research On Key Technology And System Development Of Printed Matter Defect Detection

With the rapid development of printing industry,the market is increasingly strict with the quality of printing products.The traditional manual method of detecting the defects of printed matter has some shortcomings,such as low efficiency,strong subjectivity,high rate of missed detection and poor comfort.The development of machine vision technology promotes the intelligent process of the production process,thus promoting the generation of visual detection of printed matter defects.And by virtue of its high efficiency,accuracy and objectivity,it has gradually become a research hotspot in recent years.Based on the above background,the paper takes the typical machine vision detection as the basic principle to study the defect detection technology of printed matter.According to the motion characteristics of medical flexible packaging materials in the printing process,it analyzes the imaging characteristics of linear array camera triggered by encoder.Aiming at the problems of high false alarm rate and long-time running downtime of the existing print defect detection system,and combining the requirements of the production quality and efficiency of the enterprise,it takes reducing the image noise and ensuring the stability of the system as the key technologies,and develops a print defect detection system that can be used in the production site.To solve the problems of clutter interference in the pulse waveform of the encoder and frequency division error of the image acquisition card,the paper proposes a solution of adding I/O box between the encoder and the linear array camera to filter the pulse signal of the encoder and discard the pulse function,so as to improve the image quality from the source.For the printed image with noise,first of all,the paper divides the noise into gray level noise,position noise and contour noise according to the form of noise.Secondly,based on full analysis on noise characteristics,it adopts suppression algorithms of the dynamic and static threshold,corner area-the area registration and difference of gray value in the field rows.Finally,it uses the auxiliary noise reduction methods of changing the template dynamically and based on the visual properties of human eyes.The above methods effectively reduces the influence of the three kinds of noise on the detection results,thus reducing the false alarm rate of the system.The paper monitors the image quality by calculating the image shape variable,and decides whether to carry out the initial phase registration or stop the detection and prompt the user to check the device according to the condition that the shape variable exceeds the threshold value.It represents the general process of defect detection in a modularized way.Under the restriction of producer-consumer model,it implements multithreading scheduling in the way of event object,which ensures the running sequence of each module.It monitors and adjusts the buffer capacity of each module in real time according to the image size and computer memory,so that the memory occupancy rate of the system is controlled within 95% and the stability of the system is guaranteed.In addition,in view of the serious time-consuming problem of comparison operation and Blob analysis modules,it adopts thread pool and limiting the number of defects to effectively control the running time of these two modules.As a result,the real-time performance of the detection system is improved.Referring to the existing detection software on the market,and combining the production needs of enterprises,the paper redesigns the human-computer interface of the defect detection software system.What’ more,it realizes the development of the whole detection system on the VS2017 compilation platform based on the standard of C++11,and with the help of Openv4.1.0 and MFC development library.Finally,through off-line analysis and processing of the printed image collected online,it verifies the validity of the noise reduction algorithms and the effectiveness of the methods of system stability.The detection system developed in the paper can accurately detect serious defects such as foreign bodies,ink spots and missing prints,and can also reliably detect weak defects,for example knife wire and scratches.The detection accuracy is 0.15 mm,the minimum detectable gray difference is 10,and the detection false positive rate is less than 1%.