Spectrally enhanced eddy current inspection exploiting fusion techniques

Transient eddy current techniques have been used to obtain broadband excitation. The energy however is practically concentrated at low frequencies, minimizing the information corresponding to high frequency range. An algorithm is developed to enrich the spectral information of eddy current signal, invoking image fusion techniques. C-scan images corresponding to sinusoidal and pulsed eddy current inspection are fused to increase the inspection sensitivity to identify material flaws. Principal component analysis (PCA) procedure is implemented to synthesize and reorganize information in the original spectral bands into new bands. Multi-resolution wavelet decomposition algorithms are used to identify features with different dimensions in the sample. Results based on the proposed technique, are presented using images generated by computational simulation as well as experimental inspection of samples with different flaw sizes. © 2009 American Institute of Physics