Medicine Glass Bottle Defect Detection Based on Machine Vision

Abstract

Medicine glass bottles must be inspected for various indicators after production. The article proposes a method based on machine vision for the detection of glass bottle port defects with simple operation and wide application. Obtain the image of the bottle by using the backlight illumination. Pretreat the image of the bottle, and the quality of the glass bottle is determined and the defect range is determined by detecting and analyzing the characteristics of the connected domain of the defect position of the glass bottle port. The pretreatment mainly includes image median filtering, image enhancement, and edge detection. The types of glass bottle defects mainly include cracks, missing edges, dirty bottles, black spots, and so on. After testing quite a number of medicinal glass bottles, the method can be widely applied to common glass bottle port defects. The thirty-six sample bottles with missing edge defects had the lowest recognition rate of 91.6%.