Aiming at low contrast,complex background texture,many kinds of defects and unev en brightness in the magnetic ring surface defect image,a surface defect extract ion method of magnetic ring is proposed based on an improved genetic algorithm of adaptive Canny algorithm of edge-detection met hod and masking technology in this paper.Firstly,based on the analysis of gray and gradient features magnetic surface image in different regions of the interfere nce of magneticing background area is shieldedby fitting the contour mask image in and out of magnetic ring.Secondly,an inproved adaptive Canny edge detection algorithm based on 8neighborhood anisotropic filtering is used for suppressing the interference of magnetic ring surface texture .Finally,the edge connected domain is enhanced by morphology using digital image,and the mask image is used to extract the magnetic surface defect.A large number of online experiments are carried out using the developed prototype.Experimental results show that the defect extraction algorithm have good stability,strong ro bustness, and can accurately and quickly extract the defect of various regions in magnetic ring surface image.And the accuracy rate of surface defect detection is 97.3%.