Research On Recognition Algorithm Of Beverage Bottle Quality Inspection Robot

Updating the existing beverage production line and making it intelligent is already a development trend.The mechanized production of beverage production line can not guarantee that each bottle of beverage can meet the quality requirements.Therefore,to makesure the production qualified equipe with the beverage quality inspection device in the production lines.Intelligent production lines require machine vision inspection instead of the original artificial light inspection method.At present,some beverage production lines have been equipped with the corresponding quality inspection equipment.But most of the market has been occupied by foreign equipment and few domestic manufacturers are customized to the visual company.It remains problems such as high cost and for different bottle types need to research different inspection systems.In order to meet the requirements and update the beverage production lines,this paper studies on the visual inspection system of beverage bottle and recognition algorithm.The main works and contributions of this paper are as follows:(1)A system design scheme for intelligent inspection of beverage bottle is provided.The main components and modules are introduced,and the visual imaging module and corresponding recognition algorithm are key modules for the development of beverage bottle intelligent detection equipment.(2)To solve the liquid level position recognition problem of PET bottles,a liquid level positon algorithm was designed.This paper introduces the general recognition process of liquid level,the image processing,and the gray scale projection gradient.According to the characteristics of liquid level gray projection,the projection gradient diffusion method is used to locate the liquid level position.The results of the experiment showed a certain accuracy for the position of the PET beverage bottles,providing a solution for the beverage level positioning on the beverage bottle production line.(3)To solve the bottom positioning and defect recognition problems of empty bottles,a precise positioning algorithm is designed.On the basis of Hough rough positioning,extract the exact edge points to calculate circle center with fast random circle detection algorithm.Use the bottom prior knowledge to divide the bottom of the bottle into the bottom area and the non-slip area.Finally,combined the results of the two areas as the defect recognition results and it show that the proposed algorithm has high accuracy and defect detection.(4)Apply the intelligent algorithm to the bottom defect recognition of the bottle.For the non-slip zone defect detection,the vertical gray scale projection curve of the skid zone image is used as the feature,and the BP neural network is used for training.For the whole image of the bottom of the bottle,the PCA feature of the bottom sample is selected and input into BP neural network.The basic principle of PCA and the PCA feature extraction process of bottom is introduced in detail.The experimental results show that BP neural network algorithm has a certain accuracy in the bottom defect recognition under the condition that the input feature is appropriate.In conclusion,the visual inspection system of beverage bottle quality is studied in this paper.The visual inspection imaging module and the recognition algorithm of two key problems in the system are studied.The experimental results show that the research method of beverage bottle quality visual inspection system is feasible.The research results of the defective monitoring of beverage bottles in this paper provide a way for the identification and identification of beverage bottle defects,which has important theoretical and practical significance.