Technical Report No. 2016-1 : Angle-valued Spectral Filters for Image Analysis

Fine-scale direction structure of wave fields can be valuable for assessing the structure of wind gusts in optical and radar images. We propose two new nonlinear filters that can measure the spatially-varying direction of waves in an intensity-valued image. These filters act on the local 2D Fourier Transforms of an image to obtain angle estimates. We then use simulated data to test the accuracy of these filters in obtaining spatially-varying angle fields from realistic images of ocean waves. Finally, we exhibit the behavior of our filters on SAR imagery of the ocean.