Blocked spectrum compressive sensing based on Root-MUSIC algorithm for SAR image

Abstract

In order to effectively reduce the storage volume of complex image data in high-resolution synthetic aperture radar (SAR), blocked spectrum compressive framework [1][2] based on Root-MUSIC algorithm is proposed. In this paper blockwise processing [3] of SAR image method is introduced, which can effectively reduce the storage space of measurement matrix. Gaussian random matrix is employed as observation matrix [4][5] to obtain observed value of each sub-block. Model parameters can be calculated by means of Root-MUSIC algorithm. Spectrum signal is reconstructed from small amount of measurements. Simulation results with real spatial-sparse SAR image demonstrate that data storage capacity can be reduced to as low as 1.17%, which validate the effectiveness of the method.