期刊
IET RADAR SONAR AND NAVIGATION
卷 13, 期 2, 页码 316-325出版社
INST ENGINEERING TECHNOLOGY-IET
DOI: 10.1049/iet-rsn.2018.5115
关键词
synthetic aperture radar; polynomials; computational complexity; Newton method; radar imaging; motion compensation; optimisation; accelerated translational motion compensation; contrast maximisation optimisation algorithm; inverse synthetic aperture radar imaging; parametric finite order polynomial; translational motion property; polynomial coefficient vector; image contrast; Broyden-Fletcher-Goldfarb-Shanno algorithm; quasiNewton algorithm; signal-to-noise ratio; computational complexity; pseudo Akaike information criterion; BFGS algorithm
资金
- National Natural Science Foundation of China [61771372, 61771367]
- National Science Foundation for Distinguished Young Scholars [61525105]
- Fund for Foreign Scholars in University Research and Teaching Programs (the 111 Project) [B18039]
Range alignment of traditional translational motion compensation for inverse synthetic aperture radar imaging generally cannot be implemented accurately under low signal-to-noise ratio, resulting in the following phase adjustment invalid. In this study, a novel accelerated translational motion compensation with contrast maximisation optimisation algorithm is proposed. Translational motion is first modelled as a parametric finite order polynomial. The translational motion property can be compactly expressed by a polynomial coefficient vector. Meanwhile, the image contrast is utilised to estimate the polynomial coefficient vector based on the maximum contrast optimisation, implemented by Broyden-Fletcher-Goldfarb-Shanno (BFGS) algorithm. BFGS is an effective quasi-Newton algorithm, yielding fast convergence and small computational complexity. Moreover, a method called pseudo Akaike information criterion is also proposed to determine the polynomial order adaptively. Both simulated and real data experiments are provided for a clear demonstration of the proposed algorithm.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据