A Method for Ocean Wind Direction Estimation from SAR Images Based on CONTOULET Analysis
Shamshiri, Alireza1; Keshavarz, Ahmad2
1Azad University of Bushehr, IRAN, ISLAMIC REPUBLIC OF; 2Persian Gulf University- Bushehr, IRAN, ISLAMIC REPUBLIC OF

On the basis of the SAR geometry sea waves produce in the image typical patterns, called wind streaks, aligned along the normal direction of the wind. This is the basic idea for extracting wind direction from images of the sea surface . Several works have appeared in recent literature for approaching this problem. These methods can be divided into two categories: one is in the spectral domain, the other in the spatial domain. In this paper, we propose a framework to automate wind direction retrieval based on Contourlet decomposition associated with spectral processing.

Wind vector near the ocean surface generates small scale roughness on the surface that increases with increasing wind speed. These small scale roughnesses appear as linear feature in SAR Images. In other words, if the linear features aligned with the wind direction are visible on SAR images, the wind direction can be derived directly from the images.To retrieve the orientation of these linear features, two categories of methods have been developed: spectral methods and spatial methods.

Traditionally, feature detection/extraction was done with a variety of methods, such as Laplacian operators, gradient operators, the Laplacian of Gaussians, difference of Gaussians, Canny detectors, or anisotropic diffusion. However, wavelet transforms have come into light as a means of feature detection. Wavelets are classified as a linear transform that is capable of displaying the transformed output at multiple resolutions depending on the point of time/space and at the desired frequency. In contrast to the short-time Fourier transform (STFT), the resolution changes depending on the frequency that is to be examined and at what time or spatial area is to be examined. In the 1-D case, wavelets are used for signal processing by the virtue that wavelets can store more frequency information with less coefficients and reconstruction is only limited by the coefficients that are available. Wavelets can be naively extended to the 2-D case by means of separable functions, but there is limited directional information stored in a regular 2-D wavelet transform. Because of the separability limitations, only a horizontal, vertical, and 45 degree component can be easily determined. Incidentally, edges can be seen easily, but directional information about the edge is not known. Because of this, it takes more coefficients to do a proper reconstruction of the edges. Typically, a separable 2-D wavelet transform provides:

  • Multi resolution, which is the ability to visualize the transform with varying resolution from coarse to fine.
  • Localization, which is the ability of the basis elements to be localized in both the spatial and frequency domains
  • Critical sampling, which is the ability for the basis elements to have little redundancy.
    However, it is not capable of providing:
  • Directionality, which is having basis elements defined in a variety of directions.
  • Anisotropy, which is having basis elements defined in various aspect ratios and shapes.
    There are many directional extensions of the 2-D wavelet transform that could be potentially examined that also possess directionality and anisotropy. The Contourlet transform is a discrete extension of the curvelet transform that aims to capture curves instead of points, and provides for directionality and anisotropy. Figure 1 shows the general concept of capturing curves.

    Figure 1: Conceptual visualization of curvelets/contourlets

    Contourlets are implemented by using a filter bank that decouples the multi scale and the directional decompositions. In Figure 2, Do and Vetterli [8] show a conceptual filter bank setup that shows this decoupling. We can see that multi scale decomposition is done by a Laplacian pyramid, and then a directional decomposition is done using a directional filter bank. This transform is suitable for applications involving edge detection with high curve content

    Figure 2: Filter bank for Contourlet transform.

    Here, we propose a novel approach which, in some way, tries to exploit the advantages of Contourlet Transform and Spectral processing. We use the Contourlet Transform to decompose a SAR image into Contourlet coefficients to emphasize details in different scales of the image. The Contourlet coefficients are the input to the spectral method, followed by the identification of the maximum values in the Fourier spectrum. Figure 3 shows the block diagram of proposed method.

    Figure 3: block diagram of proposed method using Contourlet Analysis and Spectral Processing.

    Briefly we extend the wavelet analysis method by using Contourlet transforms, which have the potential to improve the streak detection results.

    Wind over the ocean has not a certain direction and speed (these two factors are defined by wind gustiness and steadiness). These phenomenon cause the linear feature (created by wind) not to be exact linear (directional) and isotropic. As cited before these are two main week point of wavelet analysis which make the operator to change the scale of sub image. Now these problems are improved by Contourlet. In this paper, we have shown that the Contourlet transform can be used for feature extraction. We constructed a discrete transform that can offer a sparse representation for piecewise smooth images, as promised by the curvelet theory. There are two key features of curvelets that could lead to an improvement over the wavelet transform, namely directionality and anisotropy. The proposed method has been tested with several ENVISAT ASAR Wide Swath Images and the Results were successful for most of the images processed (about 70% ) and has more accurate results in comparison of other spectral method (like wavelets).