부산대학교 도서관

Aerial image is one of the primary data sources for underwater oceanographic studies. These images are often corrupted by clutters induced by surface water waves. Removal of the wave clutters from these images is an important preprocessing step for accurate assessment of information. In this paper, we introduce a novel technique combining the X-ray wavelet transform (XWT) with Markov random field (MRF) for directional noise removal. Surface water waves are classified according to their features into two types: ripple wave (long-wave) and spark wave (short-wave). We show in our numerical experiments that by performing XWT along the direction of wave propagation, the wave clutters can be successfully detected. To remove long-waves, resampling and subband filtering techniques are used. To remove short-waves, on the other hand, spectralspatial maximum exclusive mean (SMEM) filter is used in this study. Finally, because of the directional characteristic of the clutters, nonisotropic MRF is introduced into the post-processing step to refine the output. Experimental results show that one can remove both kinds of wave clutters with only small background distortion using the proposed hybrid algorithm. Index Terms: Declutter, Markov random field, noise removal, resampling, subband filtering, X-ray wavelet transform.