Research On Infrared Image Segmentation Method Of Aircraft Skin Damage

Aircraft skin is an important part of the fuselage.The detection and repair of damaged skin will determine the safe of the aircraft.Infrared non-destructive testing method can detect both surface and internal defects without damaging the skin.In order to identify and analyze the defects,extract the targets and build the damage database to analyze the skin damage more efficiently and accurately in the future.The primary task is image segmentation,which is of great value to quantitative analysis and engineering practice.An infrared thermal imaging detection system was established to acquire the infrared images of damaged skin specimens and reasonable methods for images enhancement and denoising were selected to pretreatment based on the purpose of the previous paragraph.Based on the traditional methods,two new infrared image segmentation methods are put forward considering the shortcomings of infrared images.In the method that based on the game between MRF and improve GVF Snake,the infrared images are firstly segmentation by OTSU,then MRF region segmentation and improved GVF Snake boundary extraction are executed independently,next,game theory is applied to fuse the results of the above two,finally find the best strategy under conflicting conditions and output the optimal segmentation when the Nash equilibrium is reached.In another method that the normalized random walk is combined with the SIR model,the mean-shift algorithm is used to improve the efficiency.And inspired by the SIR epidemic model,the random walk algorithm based on the SIR model is deduced by regarding the image pixel as bacteria,and then normalized it,the final segmentation result is obtained when the steady state of the SIR diffusion process is reached.The application of the proposed two algorithms in the infrared image segmentation of aircraft skin damage is studied.The results of each related method mentioned in this paper are analyzed and compared qualitatively.At the same time,the advantages and disadvantages of the two new methods proposed in the paper are studied comprehensively,which is valuable for image processing followed by aircraft in-situ inspection in the future.