Research On The Crack Defects Detection Algorithm Of Aircraft Skin

Every time the plane takes off and lands, its surface must endure the pressure and decompression process, which maybe make great damage to the material around rivets fixing the airplane surface, and so tiny radial cracks formed. There maybe cause unpredictable results if cracks are uncontrolled and not repaired. As a result, how to detect cracks in aging aircraft skin with dependability and efficiency has become a new research hot point. The original Magneto Optic image contains many kinds of disturbs, such as snake-like lines and the variation of the light distribution. To get binary images for recognition, motion-based filter and binarization method are used.In order to detect the aircraft surface defects, new methods based on Mathematical morphology and crack-isolated methods are presented to recognize multi-crack. Mathematical morphology is used to segment and recognize cracks, and this method's steps are: firstly, the centre and the radius of rivet areas are determined by the inner-distance mean value method; and then the crack defect is segmented by the method of self-adaptive liner star-shaped morphology erosion operator whose direction is determined by the four-point direction rapid positioning method and the centrosymmetric method is used to resolve the problem of crack defects region adhesion. Finally, the rivet regions are classified according to the existence of defect region and related characteristics. The inner-distance mean value method can detect centre exactly. And the multi-crack can segment and recognize by Mathematical morphology method.The crack-isolated method for multi-crack recognition is after image pre-processing and the determination of rivet centre, using a circle template constructed by the rivet centre and radius to cover rivet sites and then potential crack regions are isolated from the template region. Finally, the crack area is used as threshold and cracks'existence and direction are determined according to isolated regions'area and shape. The experimental results suggest that, the rivet sites can be recognized by circularity parameter. At the same time, crack area is used as threshold to judge cracks that can detect multi-crack, crack areas and crack directions.All cracks direction and length stored in database are used for damage assessment. Firstly, using Visual C++ 6.0 to establish assessment interface, and then divide aircraft skin into different regions through cluster method, finally, to assess damage through fuzzy evaluation judgment method.