The Application of Two-dimensional EMD to Extracting Wave Information from ERS SAR Images
Pingjian Song(1) and Jie Zhang(1)
First Institute Oceanography,SOA,China,
SAR(Synthetic Aperture Radar) is imaged by resonance interaction between radar wave and sea-surface micro scale wave. All marine processes that affect the distribution of micro wave spectrum can be found in SAR images, such as internal wave, mesoscale eddy, front, etc. Thus a SAR image of ocean wave often mingles with other larger scale information. In this paper, two-dimensional EMD(Empirical Mode Decomposition) method is developed, which includes key techniques of Delaunay triangulation and Bernstein-Bézier spline interpolation. This method produces a fully two dimensional decomposition of a image into a series of IMF(Intrinsic Mode Functions) and residual images with the same size as the original image. By combining IMF images, different scale information in original image is extracted. In consequence, we apply this method to ERS SAR images; wave information is successfully separated from the other information. The wave spectrums are computed both in original image and in extracted wave image, applying Hasselmann method. The results preliminary show that two-dimensional EMD provides an effective way to extract wave information from an aliasing ERS SAR image.