Study On Surface Defects Detection Method Of Permanent Magnet Ferrite Magnet Rotor Based On Machine Vision

Permanent magnetic ferrite magnet rotor is an important part of products such as submersible pumps, filters, drainage pumps, synchronous motor, etc. So, it is very critical to inspect its surface quality. Currently, the detection method for the surface quality of the permanent magnet ferrite magnet rotor is entirely dependent on the human eye in the factory, besides there are many disadvantages such as environmental influence, low efficiency, high cost, etc. if using this method. Based on this kind of situation, this paper proposes a set of defects detection method based on machine vision for the permanent magnet ferrite magnet rotor. The main contents of this paper are as follows:(1) In this paper, a bidirectional low angle lighting mode was designed, which realized the high-definition imaging of the magnet rotor surface; The algorithm of excluding "BaiTiao" of both ends of the magnet rotor was proposed, which effectively overcomed problems that the "BaiTiao" brought pseudo edge information and a large amount of data and other issues; Designed the platform mechanism for the detection of the magnet rotor surface, and analyzed the selection of industrial camera, lens and light source, at the same time, the hardware parameters of the system were set up; Through a large number of experimental comparison, it demonstrated that the hardware platform design was reasonable.(2) An algorithm based on multi-scale edge detection was proposed, which can accurately detect the edge of the target area. Firstly, we used the orthogonal wavelet transform to get the high frequency coefficients and low frequency coefficients of different scales in the image; Then, the background and the noise points were eliminated according to the local and global feature information of the high frequency coefficients in each scale; The approximate coefficient of the lowest scale was set to 0, and then the image was reconstructed; Finally, we used double threshold to deal with reconstructed image, which enhancing the edge connectivity. The experimental results showed that this method was robust to the edge detection of magnet rotor. In addition, edge thinning method based on mathematical morphology was adopted to realize the refinement of the edge of the object.(3) Analyzed the characteristics of the surface defect of the magnet rotor, and established the whole algorithm of the defect identification. In order to solve the problem of identifying the angle defect, a method based on the prior knowledge of the its position shape was proposed; For pits, cracks, scratches, convex objects and other defects, according to their morphological characteristics, gray feature, projection feature, using a principal component analysis to reduce the dimension of data, and formulated the corresponding directed acyclic graph support vector machine (SVM) algorithm to solve the problem of defect classification. Finally, the experimental results showed its effectiveness.This paper proposes a new idea of the detection method based on machine vision for permanent ferrite magnet rotor, considering the magnet rotor shapes, color, size and other priori knowledge, developed the system image acquisition platform, designed the system software algorithm and analyzed the results of the experiment. At last, a set of human-computer interaction interface was designed to demonstrate the feasibility of the system.