Research on SAR Image Reconstruction Based on Optimized Compressive Sensing Algorithm

Large amounts of high-definition SAR image data obtained by Nyquist sampling is not conducive to signal processing and transmission. Compressive Sensing algorithm adopted has effectively improved the performance of SAR imaging. Optimization for smoothed norm reconstruction (SL0) algorithm has been implemented in the paper, due to its poor norm estimation accuracy and slow convergence speed. While an approximate hyperbolic tangent function is applied to approximate norm in sparse signal reconstruction, Substitution of revised Newton direction for traditional steepest descent direction in the iteration path is adopted to accelerate convergence rate. Wavelet transform is adopted to make sparse sampling of SAR images, and measurement matrix is designed to do image compression. Then, image constructions by OPM, SP, GPRS, SL0, and optimized SL0 have been implemented. Experimental results show that optimized SL0 algorithm has the advantages of such performance Indicators as visual effects, peak signal noise ratio (PSNR) and reconstruction error over other ones.