Defects Recognition Algorithm on Magneto-Optic Image of Aging Aircraft Skin

A recognition algorithm on magneto-optic image of surface & subsurface defects in aging aircraft is proposed in this paper. The star vector method is adopted to extract the defect feature of rivet´s profile in aircraft skin. And the extracted features of rivet´s profile in aircraft skin are used as training samples to train the support vector classifier machine (SVCM) and produce a super defect classifier plane. Radial basis kernel function in the SVCM training process and grid method is selected to optimize the SVCM model. The fuzzy membership function is introduced to the algorithm to solve the wrong & refusal recognition problems in multi-type classifier. The experiment result shows that, the algorithm can detect both the existence and the direction of the defect exactly and performs a better capability of recognition