Defect Of Metal Cans Wall Based On Machine Vision

With the continuous progress of society, in the drinks industry, customers’ requirements of the food safety have become higher and higher, the competition between drinks companies has also becoming increasingly fierce. Therefore, it has become an urgent need to reduce costs of production, to improve efficiency and to ensure product quality. Among them the method of efficient metal tank inwall quality inspection is particularly important. In recent years, with the upgrading of the theoretical knowledge and hardware, the method of the metal tank inwall quality detection has increasingly developed from the artificial method to the automation and intelligent direction. And methods which are based on image processing have begun to attract more attention. The properties of metals led to the presence imaging in the detection process difficult, special shape, product features extraction problems and other issues, while metal tank detection method is less and need more research and development on it.According to the defect feature of metal cans, a defect detecting system is designed based on machine vision and digital image processing after considering the notch of metal cans jar, the defect of Hough line detection and scratches about metal cans wall, it realize that automatic detection and elimination of defects in the inner wall of the metal cans.The major contributions in this dissertation are as follows:l)The structure of the inspecting system was introduced, the hardware choice was analysised and fixed in detail, contains light source, camera, lens, gathering systems and other equipment. It selects the most suitable hardware design and ensures accurate capture images clear.2) The algorithms of whole system were studied deeply. Including gray transform, edge detection, least squares fitting circle, polar coordinate conversion and Hough line detection, etc. In particular, select most suitable the edge detection section, after Analyzing and comparing a variety of edge detection algorithm.3) Through the positioning of the regional, subregional defect detection. By locking the detection region, a substantial reduction in the useless information in the image, reducing the number of data storage. With4threads to scan and store the point defects, a large number of reducing the detection time and improve the detection speed metal tank wall.4) Successful application in the Visual Studio2010development platform, according to the particularity of the inner wall of metal tank, Theory and practice show that the metal tank inwall quality inspection method based on image processing in this study has some practical value on metal tank inwall detection problem.