Time-Frequency Analysis of SAR Image Based on Generalized S-Transform

A new time-frequency representation approach based on the generalized S-transform (GST) was developed to analysis synthetic aperture radar (SAR) images. At first, a SAR image is decomposed into one-dimensional (1-D) signals by choosing different scalable windows, rather than doing S-transform directly. Then, algorithm simulations and textural feature analysis are conducted using both basic S-transform (BST) and GST which has an adjustable time-frequency resolution. The simulation results show that the proposed method appears to work well and the GST spectrum can provide more complementary feature information than the BST spectrum for textural analysis of SAR images.