Research on aircraft skin damage identification method based on image analysis

Abstract

Abstract In the field of civil aviation maintenance, it is necessary to identify aircraft skin damage, and then formulate a maintenance plan. The traditional detection method process is more complicated. This paper proposes an image-based skin damage recognition method. After pre-processing the samples to form a unified sample library, the eigenvalues of the samples are obtained by wavelet packet decomposition and gray-level co-occurrence matrix, and finally the c and g values in the optimal RBF kernel function are selected to construct a support vector machine training model and the test results show that with a low sample size, the constructed model is more accurate in identifying normal skins and accidental impacts, and the overall system recognition rate is 81.5%.