Sparse Recovery for ISAR Imaging via Nuclear Norm Minimization

Missing entries of the backscattered data matrix deteriorate the quality of the resulting ISAR images. In this paper a data recovery method based on nuclear norm minimization (NNM) to recover the missing samples is proposed. Real and imaginary parts of the data are completed separately and a higher quality ISAR image is obtained by the 2D Fourier trans-form of the recovered matrix. The method has been compared to inexact augmented Lagrangian multipliers (IALM) and 2D smoothed LO (2D-SLO) methods both visually and quantitatively under three missing scenarios for two different missing ratios. The performance improvement is handled for both quantitative metrics such as root mean square error (RMSE) and correlation. For proposed method, RMSE scores difference can be higher than % 50 for the some extreme cases, whereas improvements in the correlation scores generally vary between %5-%10 when compared to the other methods.