Optical and sonar image classification: Wavelet packet transform vs Fourier transform

To develop a noise-insensitive texture classification algorithm for both op tical and underwater sidescan sonar images, we study the multichannel textu re classification algorithm that uses the wavelet packet transform and Four ier transform. The approach uses a multilevel dominant eigenvector estimati on algorithm and statistical distance measures to combine and select freque ncy channel features of greater discriminatory power. Consistently better p erformance of the higher level wavelet packet decompositions over those of lower levels suggests that the Fourier transform features, which may be con sidered as one of the highest possible levels of multichannel decomposition , may contain more texture information for classification than the wavelet transform features. Classification performance comparisons using a set of s ixteen Vistex texture images with several level of white noise added and tw o sets of sidescan sonar images support this conclusion. The new dominant F ourier transform features are also shown to perform much better than the tr aditional power spectrum method. (C) 2000 Academic Press.