A Corrosion Energy Feature Extract Algorithm of Aircraft Skin Based on Wavelet

The mechanism of aircraft skin corrosion is studied and a corrosion detection algorithm based on wavelet analysis is proposed. First of all, the magneto-optic image is decomposed by the wavelet analysis. In the next place a feature vector is assigned whose components represent energy in each sub-image. Lastly, a 1-nearest neighbor method classifier is applied to classify the proceeding feature vector as either corresponding to a region of corrosion or corresponding to a region of non-corrosion. The experimental results demonstrate that the proposed algorithm is insensitive to noise and the features are easy to extract. The algorithm has high recognition ratios and robustness for corrosion detection and the most important is that it can meet the real-time request.