The method for glass bottle defects detecting based on machine vision

With the increasing production of glass bottles, the detection efficiency and accuracy of glass bottles become more and more important. In order to improve the quality of the glass bottles and the detection efficiency, a method to detect glass bottles based on the connectivity domain feature was presented in this paper. This method extracts the defect features by pre-processing the collected image and threshold segmenting the image of the bottle. The method of analyzing the aperture area and width of the connected area of the bottle mouth is used to detect the image of the bottle mouth. The matching of the connected domain pixels is used to judge whether the bottle mouth is qualified and detect the defect location, and use the range of the detected defects to be compared with the range of the manually calibrated defects. The experimental results of bottle detection show that this method can accurately determine whether the bottle mouth is defective and detect the defect range of the bottle. The detection rate has reached a high level with good detection accuracy.