Effect of polarimetric information on time-frequency analysis using spaceborne SAR image

Inhalt:
The Time-frequency (TF) analysis techniques based on Fourier or Gabor transforms have been used in SAR images and adopted for discrimination and interpretation of complex targets in urban, ocean, forest fields. TF coherent and non-stationary are two important indicators computed from the TF analysis technique. This paper carries out a comparative analysis of these two indicators derived from full-pol and single-pol spaceborne SAR images in an agricultural area. From the preliminary results, with the introduction of full polarization information, the capability of targets discrimination is effectively improved in both TF coherent and non-stationary indicators even for spaceborne SAR images with very small azimuth looking angle range. The scattering mechanism will affect the level of these two indicators as one of important factors. In addition, a classification application by combining these two TF indicators and Span were carried out for full-pol SAR data and shown to provide rich information for man-made and natural targets segmentation with different types.