Experimental Study On Crack Detection Of Cylindrical Ferrite Parts Using Image Processing

Ferrite material is a kind of important non-metallic materials, which is essential to automation, instruments, computer technology and communication broadcast fields. It also has broad application prospects in the fields of aerospace and pollution treatment, in addition, the development of radio,magnetism, solid state physics and physical chemistry is closely related to the development of ferrite. Ferrite part is formed by roasting method; it is hard, brittle and easy to crack. In this thesis, magnetic powder inspection and image recognition were used to detect the cracks in the surface and near the surface, the main research contents are as follows:(1) Principle of magnetic powder inspection. Based on the analysis and research on the material properties of ferrite and the principle of magnetic powder inspection, magnetization method, magnetic media, magnetic effect time, ultraviolet lamp and camera were determined to obtain the image of the part that is attached by the magnetic powder.(2) Image preprocessing. In order to separate the crack district from the background, the image was transformed to HSI color space from RGB color space; H image and S image were used to separate the crack district. The result of color segmentation was a binary image, hole filling, image denoising, open operation, etc., were used to process this binary image.(3) Image rectification. The research object of this paper is cylindrical ferrite parts; the image taken by the camera was expanded into a rectangular plane to rectify the image. There were two steps to achieve this, firstly, getting the radius of the cylindrical ferrite part, secondly, inserting pixel values in the pixels that were added into the rectangular plane.(4) Crack recognition. According to the characteristics of crack, feature vector was created to judge the crack. A graphical user interface was made to display the results of image processing and recognition.This thesis did an experimental study about surface crack detection of cylindrical ferrite parts using image processing, it shows that the Technology and experimental scheme is feasible and the scheme can be used in industrial field with appropriate magnetic particle inspection equipment after some improvement.