Recognition of Crack Damages Based on Wavelet Transforms

Abstract

To recognize whether there exist crack damages in the inner structure of aircraft engine based on the borescope images, a method is presented by estimating the coarse feature of edges extracted from borescope images. The method first decomposes an edge into the approximation and details using wavelet transform at all of the possible scales. For the coefficients of the wavelet transform, the logarithm entropy is presented to represent the respective strength of the approximation and details, and the ratio of the two logarithm entropies is defined as coarseness factor. The coarseness factor can be used to estimate the coarseness of an edge. The greater the coarseness factor is, the coarser the edge is. Experiment results show that using coarseness factor as recognition feature of borescope images, the crack damages of the inner structure of aircraft engine can be recognized successfully.