Wavelet analysis of side scan sonar imagery for classification
Wavelet-based techniques are presented for compressing side scan sonar images communicated from UUV's through band limited communication channels. Testing with an automatrc classifier on reconstructed images demonstrates exceptional performance in preserving features required to identify mine-like objects. Wavelet domain automatic target recognition techniques are studied to further improve coefficient selection in compression. These techniques serve to enhance image fidelity in the vicinity of minelike objects while maintaining a high compression ratio.
SourceIn: High Frequency Seafloor Acoustics (SACLANTCEN Conference Proceedings CP-45), 1997, pp. 259-266