Sparse ISAR imaging using a greedy Kalman filtering approach

A Kalman filter based sparse reconstruction approach for ISAR imaging is proposed.The Greedy Kalman filter approach provides better reconstruction performance.The sparsity in the wavelet domain is exploited to improve the regional features.Image synthesis methods are presented to enhance multiple features in the image. The Compressive sensing (CS) theory provides a novel type of image reconstruction methods for radar imaging. A good image can be obtained using much less data as compared to the conventional imaging methods, however under the sparse assumptions of the scene/target in certain domains. In this paper, we present a Kalman filter based sparse reconstruction approach for ISAR imaging. As the Kalman filter has robust and excellent estimation performance in statistical settings for linear problems, it leads to good image reconstruction results for real ISAR data. In addition to the spatial sparsity of the scene, we exploit the sparsity in wavelet domain to improve the reconstruction of region-like features in the target image other than point-like features. The images obtained by assuming the sparsity in different domains are synthesized to further improve the image reconstruction. The ISAR real data processing demonstrates the performance of the Kalman filter based sparse ISAR imaging method and the effectiveness of the image synthesis methods. Display Omitted