Research On Packaging Quality Inspection System Of Penicillin Bottle Based On Machine Vision

In recent years,due to rising human costs,the industry's own characteristics and the laws and regulations issued by relevant state departments recently,The demand for machine vision inspection systems in the pharmaceutical industry is showing a blowout increasing.This article is based on the project of Packaging Quality Inspection System for Penicillin Bottle Powder Filling Production Line of Atomic Energy Research Institute,which mainly describes the penicillin bottle label detection and rubber plug gap detection.In this paper,the detection system is described in detail from selection and construction of hardware system,design and composition of the software system,development and verification of algorithm.The main contents include:Hardware system selection and construction.The hardware system of penicillin bottle label detection is mainly composed of linear array camera scanning acquisition part and executive section which consists of serial communication module and rewinder system.Penicillin bottle rubber plug gap detection hardware system includes 2D gap detection hardware system and 3D depth detection hardware system,the acquisition part of 3D depth detection is the monocular laser scanning stereoscopic imaging system.Software system design and composition.The overall building framework and interactive interface of the Qt software,multithreading technology constitute the collaborative work for image acquisition,image processing and task execution.The project data management and screening are achieved relied on MySQL database,and the communication between the software and the execution part is realized based on Qt's serial communication module.Detection algorithm development and verification.The label defect detection algorithm is a template matching algorithm for Sub-region and Grading.Comparing deviation template with max-min template,the detection accuracy,efficiency,robustness and adaptability of max-min template are all better than deviation template.The rubber plug gap detection algorithm is a comprehensive feature detection algorithm combined with 2D gap detection algorithm and 3D depth detection algorithm.The compactness of the detection area is used as the detection feature of the 2D gap detection.The 3D detection method is divided into 3Dmatching and feature detection.Because the absolute position of the bottle is randomly changed,only the feature detection method can be used.The 3D depth detection selects the deviation of the point cloud data fitting surface as the detection feature.The detection results of the rubber plug gap were obtained by these two kinds of detection methods.