The difficulty of surface defect detection of silicon steel strip under oil pollution is that the images obtained contain a variety of interference. In order to remove the interference while detecting defects, a nondestructive detection method based on image processing technology was proposed. Median filter and Wiener filter were employed to suppress the texture interference on the surface of silicon steel strip. Then the disconnected oil contour edge was connected and completed, after which oil pollution area mask was made to remove the interference of oil pollution edges. Bright spot mask was made to remove the interference of bright spot edge on oil surface. Finally, only the edge of the defect was left. Upon test, the method can effectively detect the surface defects of silicon steel strip under oil pollution interference, such as small defects, surface scratches and cracks.