Research On The Online Visual Inspection Method Of PET Bottle Packaging Quality

Defects in PET bottle packaging directly affect food safety,and it is necessary to perform on-line inspection of PET bottle packaging quality during high-speed production.This paper designs a set of PET bottle packaging quality inspection system based on machine vision,uses three cameras to shoot the cap image from different angles,and then uses matching positioning,edge extraction,contrast detection and other algorithms to process the image and achieve PET bottles.Cover defect visual inspection.The main work of this article includes the following:(1)On the basis of summarizing the status quo of industrial development of domestic and foreign academic research achievements,combined with the actual needs of the company's PET bottle packaging quality inspection,the overall scheme of the detection system was designed.The hardware system solution includes imaging design and device selection.In addition to designing organizations,the software design includes lighting modules,image acquisition modules,image analysis modules,and signal communication modules.(2)The visual inspection algorithm for PET bottles packaging defects was studied.The basic flow of the detection algorithm was designed,and the image was denoised and preprocessed using the median filter method.The bottle cap was positioned using the gray correlation matching method,and this was used as a benchmark to perform secondary detection for each detection subarea.Positioning,and extracting the SUSAN operator edge of each area as the main features,and then using preset rules to achieve the detection of defects such as no cover,high cover and skew cover.The Bayesian estimation method is used to automatically obtain the detection parameter threshold in the detection of the no cover defect,so that the system has certain intelligent ability.(3)Based on the first two parts of the work,the design and implementation of PET bottle packaging defect detection software,including software architecture design,human-computer interaction design and implementation,the specific designand implementation of detection algorithms,system integration and other work.(4)The test system completed in this paper was verified by actual experiments:2000 bottles of PET bottles were tested.The system detection speed can reach 15bottles/sec,and the detection accuracy is higher than 99.30%.The experimental results show that the system and algorithm designed in this paper can meet the actual production requirements.