Research And Design Of Metal Label Defect Visual Inspection System Based On Android

With the rapid development of China's packaging industry and the continuous improvement of modern testing technology,the automatic detection of surface defects in the field of packaging processing has gradually become the focus of the industry development.Among them,machine vision has become a key technology to promote automation and intelligent upgrade of the industry due to its advantages of precision,high efficiency and easy integration.At present,metal labels are attached to the surface of packaging,and most companies still use manual labeling and testing.In order to reduce labor costs and improve detection efficiency,companies hope to introduce machine vision defect detection technology.The traditional PCbased machine vision system is costly to build and is difficult for enterprises to accept.Aiming at this problem,this paper designs a visual detection system for metal label defects based on Android.The Android smart terminal is used to collect the metal label image of the packaging surface entering the shooting area.After the defect detection software processes,the detection result is sent to the PLC to control.The conveyor completes the sorting of the package.The research work of this paper includes the following aspects:1.The common defects of the metal label attached to the surface of the rigid packaging box were analyzed.According to the common defect characteristics of the packaging surface and the production characteristics of the process,the overall design was carried out,and the software and hardware system architecture design and the selection of major hardware equipment were completed.2.According to the accuracy requirements of the label defect detection system,the Brenner gradient function is used to complete the auxiliary focus before the camera calibration,and then the camera calibration is completed using a calibration tool based on the Zhang calibration method.Through experimental tests,the camera image recognition accuracy reaches 0.04 mm,which meets the design requirements.3.The three feature point extraction algorithms and two feature point matching algorithms are experimentally studied.The Lowe's and RANSAC algorithms are further used to improve the image registration accuracy and robustness,and the ORB/BF combination algorithm with better comprehensive performance is determined.Fast and accurate registration.Finally,the image difference method is used to obtain the defect pixel distribution,and a set of lightweight algorithm for Android platform is obtained.4.Built a defect detection system based on Android and completed the development of defect detection software.In the test of common defects of metal labels,the average detection rate of the system is over 90%.The validity and practicability of the defect detection system designed in this paper is verified.The laboratory test results show that the developed Android-based metal label defect visual inspection system can more accurately identify metal label defects.Stable operation and low cost,basically meet the application needs of enterprises.