The Research On Beer Bottle Mouth Defect Inspection System

In order to reduce the cost，a lot of beer and beverage production enterprises oftenrepeated use filling bottles. However, recycling bottles will have different degree of wearand tear, whose quality can not get guaranteed effectively. And more, it is very difficult touse foreign empty bottle inspection system to reach our rated technical indicators. Ourcountry recycling bottle inspection system is still in the research phase; therefore, domesticenterprises have not applied overall production of recycling bottle detection system.Thistopic is from the research of the talents to deepen teaching plan-Backbone teachers of theBeijing municipal education commission “The Research on Beer Bottle DamageInspection System ".This test system is made up of image acquisition system and image processing system.After selected reasonable light, the beer bottle images can be gained by the CCD camera;And it can test bottle defect status by various algorithms processing, then the high-speedempty bottle detection system can be formed. The image processing includes severalaspects: image pre-preprocessing, bottle mouth positioning, feature vector extraction anddamage identification and so on. In the part of image pre-processing, this paper finds thethreshold of image through the Otsu algorithm, then makes morphological imageprocessing for images, chooses Gauss filter for the image smoothing; In the part of bottlemouth positioning, locating the bottle is based on Hough transform method, then the centerof bottles can be found out; In the part of feature vector extraction and damageidentification, the paper puts forward based on analyzing connected region and momentinvariant to extract characteristic vectors, the weight matrix and threshold matrix can beobtained by using the Back Propagation neural network training samples, Which are usedfor bottle mouth defect identification. Based on a large number of experiments, therecognition effects are more ideal. this method not only meets the real-time requirement,and also prove ensures accuracy of inspection results.