Design Of Full Bottle Inspection System Based On Machine Vision

With the continuous expansion of the beverage market,the degree of automation of the production lines in beverage production plant is also increasing,and the production efficiency is getting higher too.However,it may produce some unqualified products in the production process,including capping unqualified,filling unqualified,printing unqualified,these defects are generated in the filling process and after the filling process,so the inspection of capping,filling and printing also called full bottle inspection.The traditional method of manual inspection is inefficient and poor in stability,and has been unable to meet the production needs of manufacturer.There is an urgent need for a system that enables automatic inspection of beverage products quality.This paper presents a full bottle inspection system based on machine vision,which utilizes advanced image processing technology to realize the automatic inspection of filling level,capping quality and printing quality.According to the characteristics of the beverage production line and the functional requirements of the inspection system,the overall scheme of the full bottle inspection system was designed.The inspection system is divided into hardware part and software part.The hardware part mainly describes the selection and design of each hardware module,including:(1)Through the camera,light source and image acquisition card to realize the image acquisition and transmission;(2)Use the industrial computer to complete the image analysis and processing;(3)Take the PLC and fault rejection as defective removal device.The software part mainly analyzes the specific design of the inspection algorithms and the software programming of system functions.The inspection algorithms are divided into the printing inspection algorithm,the cap inspection algorithm and the fill level inspection algorithm.In the part of the printing inspection algorithm,the Otsu threshold segmentation method is used to locate the printing area automatically.Furthermore,the threshold segmentation combination morphological processing is used to segment characters.Last,the LM algorithm based improved BP neural network is used to train and recognize the characters,and then find the missing and erroneous characters is easy.In the part of the cap inspection algorithm,the Harris corner detection algorithm is used to fit the straight line of the embryo ring.Furthermore,the Canny algorithm and the least squares algorithm are used to fit the straight line of upper edge of the cap.Taking the straight line of the embryo ring as a reference line,and the capping quality can be obtained through calculate the angle and distance between the line of upper edge of the cap and the reference line.In the part of the fill level inspection algorithm,a fill level inspection method based on pixel statistics is proposed,that can judge whether the filling capacity is qualified or not by calculating the number of target pixel after threshold processing.After realizing the above inspection algorithms,we design the software system to integrate these algorithms,and program the corresponding function module.Finally,the whole inspection system is accomplished.This paper designs a full bottle inspection system based on machine vision to realize the automation inspection of PET bottle's printing quality,capping quality and filling quality.Experimental result shows that the accuracy of the system inspection is above 99%and can inspect 36,000 bottles per hour.The system can replace the traditional manual inspection and improve the accuracy and speed of full bottle inspection,and has high engineering application value.