The Design And Implementation Of Printed Defect Detection System Based On FPGA

The manufacturing industry is the manifestation of the national economic strength.Printing industry as a part of it,facing the problem that how to use the advanced technology to innovate and develop itself with the rapid development of the technology.In the actual production process,print quality problems occurred frequently,such as missing prints,ink dots,scratches and other defects,which will also affect subsequent printing,directly determining whether the printed materials can be recognized and satisfied by customers.However,in the past,the detection and identification of the quality problems of printed products were all observed by the human's eyes,which took a long time but with low accuracy.Therefore,in order to solve these defects,it is very important to design and develop a defect detection system based on FPGA(Field Programmable Logic Gate Array),which is capable of performing efficient and real-time detection of printed product defects,to solve these problems in the current printing industry.The development process of this system has undergone on-the-spot investigation,and its functions are in line with the actual application,which can solve the printing defects during the production process satisfactorily.This system builds a visual Web system of human-computer interaction.First,after taking photos for the printed products on hardware platform,the user can submit a standard image template and a detection process by interaction module.After uploading the data to the background,the Web system uses the template to match the images to obtain the detection area,which will be tested for different types of errors,and the detection result will contain two parameters:the type of error and the location of error.The system will mark the corresponding the location of error and indicate the the type of error,and feedback the test results in real time to facilitate customers to deal with the specific type of error.This Web system is written by Python language,using the latest Vue.js front framework to design and develop,using modular programming to facilitate secondary developmentThe image segmentation module of this system adopts FPGA to realize acceleration,so that the image segmentation process can be completed at high speed In the classification and detection program of printed products,HOG(Histogram of Oriented Gradients)feature extraction and SVM(Support Vector Machine)classification methods are used,so that the system can effectively identify and classify different printed products'defects.At the same time,relevant interface of image processing of OpenCV is used to realize the specific implementationsBased on machine vision,this system has effectively solved the existing detection defects of printed products on the market such as low interactivity,low speed,high price and low accuracy,etc.Compared with the traditional print detection system,it has made great progress and promoted the development of the entire print industry.