Robust and efficient SAR image coding transmission based on compressive sensing

In this work, a new robust and efficient airborne synthetic aperture radar (SAR) image coding transmission scheme based on compressive sensing (CS) against lossy channels is proposed. The robustness is achieved using the democracy of CS. Considering the poor R-D performance of the traditional CS due to SAR image's weak sparsity, we use directional lifting wavelet transform (DLWT) as sparse representation and sparse-filtering to eliminate the interference of small coefficients. By exploiting the inter-scale dependency of DLWT coefficients, an efficient Bayesian reconstruction algorithm is adopted. Furthermore, optimal tradeoff between bit-depth and measurement rate is used. Experimental results show that the proposed scheme is more robust against packet loss compared with the traditional joint source-channel coding (JSCC) scheme. When the packet loss rate (PLR) is excessive, the JSCC scheme easily leads to cliff effect, however, the R-D performance of the proposed scheme decreases more gracefully while achieving a comparative R-D performance.