Aircraft Skin Damage Detection Method Based On Machine Vision

Owing to the extreme operating circumstances of aircraft, besides the fatigue damage caused bynormal flight, there are problems such as structure corrosion which is caused by environment factorand so on. These defects will cause damage to the structure integrity of aircraft skin, thus leading tothe decrease in the strength of skin structure. During the operation of aircraft, in servicenondestructive testing is very important. According to the design requirements of damage tolerance,the defects should be checked out before the structural strength decreases to residual structuralstrength and be repaired to restore to the structural design strength of aircraft. At present, the aircraftskin detection is mainly based on manual inspection, and there are labor-intensive, long test cycle,high rate of undetected problems, etc. How to realize the efficient and accurate detection of aircraftskin damage is becoming one of focuses and hot issues in the skin damage detection research field.First of all, a recognition system based on the machine vision has been designed specifically forthe detection of aircraft skin damage. The system is mainly consist of image acquisition module,wireless communication module, real-time image processing module, and on-line classification andrecognition module. The image acquisition module uses the type DV-905H area-array CCD camera ofDEVELE company. The wireless communication module which is made up of transmitting end andreceiving end realizes the wireless transmission of digital image. The real-time image processingmodule is consist of the steps such as image pre-process(image filtering and image segmentation) andcharacteristics extraction(corrosion characteristics extraction of aircraft skin based on the rivetneighborhood and texture characteristics extraction of skin based on gray level co-occurrence matrix).Through the on-line classification and recognition module, on-line classification and extraction ofskin damage information has been realized.Secondly, in the phase of characteristics extraction, because the rivet neighborhoods areimportant feature areas with regard to aircraft skin, the detection of aircraft skin damage can berealized through the characteristics analyzing of corrosive neighborhood of rivet. An aircraft skincorrosion characteristics extraction method based on rivet neighborhood is proposed. The methodimproves the algorithm which is based on the inner-distance mean value method for determining thecenter and the radius of rivet areas, reducing the center deviation degree of corrosion rivet andrealizing locating the center of rivet accurately.Then, an aircraft skin texture characteristics extraction algorithm which is based on the gray level co-occurrence matrix is proposed specifically at aircraft skin texture characteristics. Based onestablishing the image sample library of aircraft skin damage, texture characteristics extraction ofdifferent types of aircraft skin is achieved by extracting five characteristics with better effects in graylevel co-occurrence matrix which are angular second moment, inertia moment, correlation, entropyand inverse difference moment.Finally, By using support vector machine to realize the classification and recognition of aircraftskin image, and by integrating the support vector w and the hyper-planes intercept b under thecompilation environment of MATLAB and VC++, an on-line detecting software for aircraft skindamage has been developed to realize the on-line classification and detection of aircraft skin damageinformation intuitively, efficiently and accurately.This research is of great significance for improving the level of detection of aircraft skin damage,shortening the maintenance time of aircraft, and ensuring the safety of flight.