Research On Infrared Image Preprocessing Method Of Aircraft Skin Damage

Aircraft skin is an important part of aircraft and one of the key factors to determine whether the aircraft can fly safely.Therefore,it is very necessary to detect and deal with the damaged aircraft skin.Infrared nondestructive testing can detect the surface damage and internal defects of aircraft skin without destroying it.In order to better identify and analyze aircraft skin damage and defects,and effectively extract target features and provide basis for subsequent implementation of fault diagnosis,the primary task is to preprocess the infrared image of aircraft skin damage.Based on the above purposes,firstly,an infrared thermal imaging detection system is established to collect infrared images of damaged aircraft skin.Secondly,the infrared image denoising is studied based on compressed sensing theory.According to the characteristics of infrared image and noise,two improved infrared image denoising methods are proposed,which are to optimize the observation matrix in compressed sensing theory and to improve the reconstruction algorithm in compressed sensing theory.The observation matrix is optimized which considers these two aspects: the independence of the observation matrix column and the correlation between the observation matrix and the sparse base.The optimized observation matrix is used to achieve better noise removal of infrared image.The improved reconstruction algorithm is improved by introducing the similarity matching criterion of generalized Jaccard coefficients to select the best atom.The possibility of selecting the correct candidate set is improved by the generalized orthogonal matching pursuit algorithm of multiple candidate sets,and the reconstruction accuracy of the algorithm is further improved.Finally,the two improved algorithms are applied to denoise the infrared image of aircraft skin damage,and the simulation experiment is carried out by MATLAB.The simulation results show that the proposed algorithms have faster convergence speed,higher peak signal-to-noise ratio and better subjective effect than the traditional algorithms.From the subjective and objective analysis,the proposed algorithms can achieve better denoising effect.