sparse bayesian learning in ISAR tomography imaging
来源期刊：中南大学学报(英文版)2015年第5期
论文作者：Su Wu-ge WANG Hong-qiang Deng Bin WANG Rui-jun QIN Yu-liang
文章页码：1790 - 1800
Key words：inverse synthetic aperture radar (ISAR); tomography; computer aided tomography (CT) imaging; sparse recover; compress sensing (CS); sparse bayesian learning (SBL)
Abstract: Inverse synthetic aperture radar (ISAR) imaging can be regarded as a narrow-band version of the computer aided tomography (CT). The traditional CT imaging algorithms for ISAR, including the polar format algorithm (PFA) and the convolution back projection algorithm (CBP), usually suffer from the problem of the high sidelobe and the low resolution. the ISAR tomography image reconstruction within a sparse Bayesian framework is concerned. Firstly, the sparse ISAR tomography imaging model is established in light of the CT imaging theory. Then, by using the compressed sensing (CS) principle, a high resolution ISAR image can be achieved with limited number of pulses. Since the performance of existing CS-based ISAR imaging algorithms is sensitive to the user parameter, this makes the existing algorithms inconvenient to be used in practice. It is well known that the Bayesian formalism of recover algorithm named sparse Bayesian learning (SBL) acts as an effective tool in regression and classification, which uses an efficient expectation maximization procedure to estimate the necessary parameters, and retains a preferable property of the l0-norm diversity measure. Motivated by that, a fully automated ISAR tomography imaging algorithm based on SBL is proposed. Experimental results based on simulated and electromagnetic (EM) data illustrate the effectiveness and the superiority of the proposed algorithm over the existing algorithms.
Su Wu-ge(苏伍各), WANG Hong-qiang(王宏强), Deng Bin(邓彬), WANG Rui-jun(王瑞君), QIN Yu-liang(秦玉亮)
(School of Electronic Science and Engineering, National University of Defense Technology, Changsha 410073, China)
Abstract:Inverse synthetic aperture radar (ISAR) imaging can be regarded as a narrow-band version of the computer aided tomography (CT). The traditional CT imaging algorithms for ISAR, including the polar format algorithm (PFA) and the convolution back projection algorithm (CBP), usually suffer from the problem of the high sidelobe and the low resolution. the ISAR tomography image reconstruction within a sparse Bayesian framework is concerned. Firstly, the sparse ISAR tomography imaging model is established in light of the CT imaging theory. Then, by using the compressed sensing (CS) principle, a high resolution ISAR image can be achieved with limited number of pulses. Since the performance of existing CS-based ISAR imaging algorithms is sensitive to the user parameter, this makes the existing algorithms inconvenient to be used in practice. It is well known that the Bayesian formalism of recover algorithm named sparse Bayesian learning (SBL) acts as an effective tool in regression and classification, which uses an efficient expectation maximization procedure to estimate the necessary parameters, and retains a preferable property of the l0-norm diversity measure. Motivated by that, a fully automated ISAR tomography imaging algorithm based on SBL is proposed. Experimental results based on simulated and electromagnetic (EM) data illustrate the effectiveness and the superiority of the proposed algorithm over the existing algorithms.
Key words:inverse synthetic aperture radar (ISAR); tomography; computer aided tomography (CT) imaging; sparse recover; compress sensing (CS); sparse bayesian learning (SBL)
