On-line Inspection System For Bottle Mouth Defects Of Liquor Bottle Based On Machine Vision

In the production process of the bottles of liguor,due to the limitation of production technology,there exist some defects at the mouth of bottle.When the bottles,which have those defects,used by the consumer,some of them can affect the appearance of products,and serious ones may lead to potential safety hazards.In order to meet the needs of liquor bottle manufacturers,an online detection system which is based on machine vision for the defects of the liquor bottle mouth is proposed to realize the automatic detection of bottle mouth defects that affecting sealing or appearance.This thesis designs the system of liquor bottle mouth defect detection,including imaging mechanism,image processing module and motion control module.After a large number of imaging experiments,the selection of camera,lens and light source is determined,and a imaging mechanism is constructed.The image processing algorithm is compiled.Firstly,the bottle mouth region is located by fitting the circle of the outer edge of the bottle mouth.Secondly,the algorithm of extracting the nearest circle is proposed to realize the rough location of the bottle mouth end surface.Thirdly,the precise location of the bottle mouth end surface is achieved by combining threshold segmentation with ellipse fitting.Then,using the polar coordinates to expand the bottle mouth end image,and applying the hysteresis threshold segmentation method to extract the defect area.At last,defects are detected according to the character of the height.The motion control module is designed to simulate the conveyor belt of the liquor bottle production line with slide rail.Photoelectric sensor is used as external trigger signal source to control the camera to collect images,and PLC is used for signal transmission.Write a liquor bottle mouth defect detection software interface to visualize defect detection for user convenience.The algorithm was tested with 712 images in the gallery.There was no missing detection of defective bottle mouth,but there were 30 false detections of the good bottle mouth image.The false detection rate is 5.51%.The reason of false detections is the defect of bottleneck and the rough and inferior bottle mouth.89 liquor bottle samples were used to test the whole system.The system runs well.There are no exception throwing,no missed detection and 3 false detections.According to the calculation,the detection resolution of the device is twice that of the human eye,able to meet the needs of liquor bottle manufacturers.It has been proved that the online detection system of bottle mouth defects of liquor bottles designed in this thesis is stable in operation and has high detection accuracy,and has practical application value.