Research On Surface Defects Visual Inspection System For Metal Components

Under the circumstances of industrial mass production, the traditional artificial traditional artificial detection can’t satisfy the requirements of the modern industrial quality inspection any longer for its defects, such as low speed, vulnerable to be affected by workers and so on, moreover, the number of researches on the detection which are based on machine vision is increasing. However, the metal components always along with features of little volume and complex shape, so the detection on its surface is often different with the plate, and there are few studies and applications of it. Meanwhile, to detect a new part means to develop a new detection system, which is a waste of resources. In terms of these problems, the paper designs and implements the surface defects visual detection system of metal components.Requirements analysis of the system is explained in the paper, and the structures of software and hardware are designed according to the requirements. The detection of components is completed by the cooperation of the software and hardware of the system, among which, the SCM controls the behavior of the components of machines, while the IPC controls the lighting source and the collecting card to coordinate the behavior and collect pictures of components by communicating with the SCM, all pictures will be detected by the software on the IPC.Analysis of the lighting scheme is illustrated, after describing various theories of light modes and applied occasions, a collecting method which combines the techniques of multi-directions, multi-cameras, bright and dark field and a component of multi-shooting is proposed.According to the metal component’s character of complex shapes, an extraction plan about the region of interest which is based on the prior knowledge and the configuration modeling is advanced. The prior knowledge is helpful to reduce the computation and difficulty of matching location of the region of interest, while combing the configuration modeling with object-oriented technology and XML files is benefit for the system to detect components with complex shapes in good scalability and adaptability.An operating mechanism based on the configuration which is easy to parallel and can automatic remove the repeated computations is put forward by analyzing the detection process. The system can parallel execute the detection algorithms by configuring the algorithm, the algorithm that eliminates the repeated computations can let the system to optimize the detection algorithms automatically, while, as a component, the detection algorithm should be extended to be applied to some unknown flaws.The surface defects visual detection system of metal components is implemented in this paper, and by applying it onto some samples proves that the system to be feasible. Thirty figures, three tables, fifty-three references.