Abstract: Defects may appear on the surface of the magnet during cutting into smaller slice. These defects will seriously affect the performance and service life of magnetic products. So automatic surface cutting defect detection becomes an important task for magnet producing. A Fourier image reconstruction based magnet surface image process method is proposed, to the question that defect of magnet surface with dark color and low contrast can not be extracted via the classical defect detection algorithm. Fourier transform is used to get the spectrum image of the magnet image, and in the spectrum image, the defect will be shown as a bright line. Hough transform is used to detect the angle of bright line, then the frequency component of this line is removed, and revise Fourier transform is applied to get the background gray image. In the end, the defect region can be obtained by evaluating the gray-level differences between original image and reconstructed image. Experiments show that the method proposed can detect surface cutting defect of magnet accurately and efficiently.

Key words: defect detection, image process, magnet, Fourier transform

摘要： 磁片在被切割为更小的片时表面上可能会出现缺陷，这些缺陷将严重影响磁性材料产品的性能和使用寿命，因此表面刀纹缺陷自动检测成为磁片生产中一个重要的任务。针对经典缺陷检测算法不能很好地提取颜色暗、对比度低的磁片图像的缺陷的问题，提出一种基于傅里叶变换重建图像的磁片表面图像处理方法。用傅里叶变换获取磁片图像的频谱图像，缺陷在频谱图像中被显示为一条亮线。用霍夫变换检测亮线的角度，去除这条亮线的频率分量，使用傅里叶反变换得到去除掉缺陷的正常灰度图像。缺陷区域则可以通过评估原始图像和重建图像之间的灰度差来获得。对大量的磁片图像进行实验后表明，该方法可以准确、高效地检测磁片表面的刀纹缺陷。

关键词: 缺陷检测, 图像处理, 磁片, 傅里叶变换