Aircraft Rivets Defect Recognition Method Based on Magneto-optical Images

With magneto-optical images for the aircraft rivets, a new automated recognition algorithm of inspecting the existence and the direction of cracks based on fuzzy support vector machine (FSVM) is presented. The binary image of rivet is obtained by preprocessing the magneto-optic image, the star radial vector method is used to acquire the feature of rivet edge, and the approximate center of rivet is got according to the threshold method. The kernel parameter and the penalty content are optimized by using the grid method, and FSVM is adopted to avoid the refusal classification and the false classification in multi-classifier. The inspection effect and the real-time performance are proved by these experiments.