Scaled Radon-Wigner Transform Imaging and Scaling of Maneuvering Target

Cross‐range scaling method of inverse synthetic aperture radar image based on discrete polynomial‐phase transform

“…Extensive algorithms have been proposed for estimating the second-order phase coefficients, for example, imagecontrast-based search [17][18][19] and scaled RWT [20]. All these methods are based on the maximum-likelihood estimation.…”

“…Considering the sampled number N, the spectrum resolution of (15) is 2π/N which is the estimation accuracy of ω m (t). Substituting the spectrum resolution and the optimal value of t as N/2 [21] into (18), we can obtain (20) where the subscript m is omitted. In the CPI, the bandwidth induced by the Δa 2 is Δa 2 NPRI/π, which equals to 2PRF/N after further simplification.…”

“…For ISAR targets with uniform motion, the rotating angle can be obtained via the rotating angular velocity multiplied by the CPI. Extensive algorithms on the ISAR images scaling have been proposed [9][10][11][12][13][14][15][16][17][18][19][20]. For multi-static ISAR, several methods based on dominant scatters [9], randomised Hough transform [10] and time-frequency distribution analysis and CLEAN technique [11] have been proposed for the interferometric ISAR (In-ISAR) image scaling.…”

“…Based on the relationship between the second-order phase coefficients of range echoes and the range geometry, the rotating angular velocity can be estimated through the least-square-error (LSE) method after acquiring the second-order phase coefficients of range echoes [17][18][19][20]. Image contrast maximisation-based methods [17][18][19] and scaled Radon Wigner transform (RWT) [20] are exploited to estimate the second-order phase coefficients. However, to achieve high estimation accuracy, small search interval must be used which will induce heavy computation load.…”

“…Motivated by the methods proposed in [17][18][19][20], a novel method based on the second DPT and LSE is proposed to estimate the rotating angular velocity and achieve the following cross-range scaling. The proposed method can achieve a better tradeoff between the precision and the computation burden.…”