Research On The Application Of Image Processing Technology In The Printing Defect Detection

Automation and intelligent is an important development direction of the modern industry, but the current print defect detection technology in our country is still immature and imperfect. For these reasons, from the perspective of improving the real-time and the applicability of the print defect detection, this paper mainly researched the variety of image processing algorithms and its application in printing defect detection.First of all, the various image preprocessing methods were researched and compared, such as color image grayscale, image filtering technology, etc. From the point of view of the processing effect and real-time, this paper selected the suitable processing algorithms for printing defect detection to enhance the quality of images and make the image more convenient for the subsequent processing.Secondly, the image registration technology was studied. For the off-line detection, the PV interpolation method and the improved direction acceleration optimization search strategy were used to determine the maximum mutual information value of the image. This method can realize the registration of the template image and the detected image.In this paper, the methods of image difference,image binaryzation and mathematical morphology were used to detect the defects of printing.Then, this paper focuses on the defect recognition of printings.The method of BP neural network was used to identify the printing defects and a new method of printing defect recognition based on convolutional neural network was proposed. This method can directly use two-dimensional image as the input, by training a large number of sample data to learn the essential characteristics of the image. At the same time, this paper collected samples, made the data set and designed the corresponding neural network. Through the experimental analysis and comparison, the advantages and disadvantages of these two methods were obtained.At last, this paper designed and developed a software system for printing defectdetection using VC6.0 and OpenCV. Its main function modules include system login,defect detection, information management, system management, customer inquiries.And it has a certain practical value.