Visual Detection Of Small Permanent Magnet Devices Billet Surface Quality Study

In the production process of NeFeB magnetic component, in view of the present status that the dimension measurement and surface defect inspection of the tiny magnetic billet is realized by manpower, the optic- mechatronics integration and computer vision technology are applied in this paper to present the visual inspection system. Firstly analyzing the inspection demand and the feasibility of non-contact detection; then summarizing the previous research situation about surface characteristics testing techniques, so as to tease out this technical solution. In this paper an overall project is designed for the visual inspection system about the and surface defect of tiny magnetic billet, which consists of orderly convey mechanism for distinguishing the scrap, image acquisition, visual inspection,analysis and judgment subsystem, separation institutions and coordinate various part of control subsystems etc. The system makes for the orderly transfer, the image automatic acquisition, dimension measurement and surface defect recognition and sorting to realize the organic integration. In addition comprehensively analyzed characteristics of lighting source, optical lenses, camera and image acquisition card and so on, so as to adopt appropriate illuminant, which meets detection requirements, lens magnification and camera pixels, to ensure the precision of subsequent analysis and measurement calculating.The surface features testing can be parted the edge detection and defect recognition detection. The detection effect can be impacted by hardware performance and its matching parameters, besides the most critical factors are rationality and calculation precision, etc, which comes from the algorithm about the image processing and edge detection and identification for online acquisition images. According to the detection accuracy of production field, firstly to preprocess the image, such as removing the noise, shadow elimination, image enhancement and so on. Based on comparatively analysis histogram method, information entropy method and OTSU method the first approach was used to realize the target and background separation. In edge detection aspect, after comprehensive comparison on the classic detection operators, such as Roberts operator, Log operator, the mathematical morphological method was put forward. A kind of improved CHT algorithm is described, which is used for pixel identification of the image edge and to calculate the graphical geometric elements parameters, which can be transformed through system calibration. This paper constructs a search algorithm based on single connected enclosed—area to recognize the surface defect; Through calculating the characteristic parameters, such as area, perimeter and circularity of defect area to establish the classification rules so as to distinguish the scrap of tiny magnetic billets. The edge inspection and identification algorithms are programmed in the MATLAB7.6 environment to realize. Finally using VC++ language to compile the procedures of image acquisition, pretreatment, segmentation, edge detection and graphical recognition, size,defects calculation of tiny magnetic billets, also to format the relevant interfaces, for instance the whole system control and system parameter setting. Of which the software integrate as one formed this visual inspection system for size measurement and surface defect of tiny magnetic billet. Finally, the inspection system is applied to a group of magnetic billets provided by manufacturers for the experiment, the result of which shows that the visual inspection system is doable on dimension measurement and surface defect inspection. The algorithms in this paper are simple but robust, can satisfy the practical detection precision, accuracy and real-time requirements. The visual inspection system designed can replace manpower in production field to realize a non-contact detection, the stability and precision both can meet the need of factory.