-

Öz

ISAR imaging based on the 2D linear prediction uses the l2norm minimization of the prediction error to obtain 2D AR model coefficients. However, this approach causes many spurious peaks in the resulting image. SVD truncation of AR coefficients depends on the choice of scattering coefficients and a wrong choice may cause underestimation of scattering centers or inefficient suppression of sidelobes. In this study, we present sparsity regularized AR models and apply them to the problem of high resolution radar imaging. By using the sparsity prior we constrain AR coefficient vector to be sparse. The use of resulting coefficient vector yields radar images with reduced side lobes improving the discrimination of the target from the background. This method also works successfully in case of narrow frequency band and angular sector. The proposed sparse AR models have been applied to the ISAR imaging problem as well as classification of ISAR images. The results show that the proposed method has higher performance compared to the other AR based methods

Anahtar Kelimeler

• Radar imaging, linear prediction, autoregressive modeling, sparse representations, regülarizasyon

Seyreklik Güdümlü Doğrusal Öngörü ile Yüksek Çözünürlüklü Radar Görüntüleme / High Resolution Radar Imaging with Sparsity Driven Linear Prediction

Öz

Özet2B doğrusal öngörü temelli ISAR görüntüleme 2B AR katsayıların eldesi için 𝑙2 norm minimizasyonunu kullanır. Fakat bu yöntem sonuç görüntüde yalancı tepelerin oluşmasına neden olur. AR katsayılarına TDA kesmesinin uygulanmasının başarımı saçıcı sayısının kestirimine bağlıdır. Saçıcı sayısının yanlış kestirimi bazı saçıcıların kestirilememesine ya da yan lobların etkin şekilde indirgenememesine neden olur. Bu çalışmada, seyreklik regülarizasyonlu AR modeller sunulmuş ve yüksek çözünürlüklü radar görüntüleme problemine uygulanmıştır. Seyreklik öncelinin kullanılmasıyla AR katsayı vektörü seyrek olmaya zorlanmıştır. Elde edilen seyrek katsayı vektörünün kullanılmasıyla hedefin geri plandan daha kolay ayırt edilmesine olanak veren yan lobları indirgenmiş radar görüntüleri elde edilmiştir. Önerilen yöntem dar band-dar açı durumunda da başarıyla çalışmaktadır. Önerilen seyrek AR modeller radar görüntülemenin yanısıra ISAR görüntülerin sınıflandırılmasına da uygulanmıştır. Sonuçlar önerilen yöntemin diğer AR temelli yöntemlere göre daha yüksek başarıma sahip olduğunu göstermektedir

 

Abstract

ISAR imaging based on the 2D linear prediction uses the 𝑙2 norm minimization of the prediction error to obtain 2D AR model coefficients. However, this approach causes many spurious peaks in the resulting image. SVD truncation of AR coefficients depends on the choice of scattering coefficients and a wrong choice may cause underestimation of scattering centers or inefficient suppression of sidelobes. In this study, we present sparsity regularized AR models and apply them to the problem of high resolution radar imaging. By using the sparsity prior we constrain AR coefficient vector to be sparse. The use of resulting coefficient vector yields radar images with reduced side lobes improving the discrimination of the target from the background. This method also works successfully in case of narrow frequency band and angular sector. The proposed sparse AR models have been applied to the ISAR imaging problem as well as classification of ISAR images. The results show that the proposed method has higher performance compared to the other AR based methods.

Anahtar Kelimeler

• Radar görüntüleme, dogrusal öngörü, özbağlanımlı modelleme, seyrek gösterilimler, regularization

Kaynakça

1. Özdemir, C., Inverse Synthetic Aperture Radar Imaging with MATLAB Algorithms. New Jersey: Wiley, 2012.
2. Gupta I.J., “High resolution radar imaging using 2-D linear prediction”, IEEE Transactions on Antennas and Propagation, Vol. 22, 31-37, Jan. 1994.
3. K.T. Kim, S.W. Kim, H.T. Kim, “Two dimensional ISAR imaging using full polarization and superresolution processing techniques”, IEE Proceedings Radar, Sonar and Navigation, Vol. 145, 240-246, Aug. 1998.
4. I. Erer and A.H. Kayran, "Superresolution ISAR Imaging Using 2-D Autoregressive Lattice Filters", Microwave and Optical Technology Letters, 32, 81-85, (2002).
5. Odendaal, J. W., Bernard, E. ve Pistorius, C. W. I. (1994). Two-Dimensional superresolution radar imaging using the MUSIC algorithm, IEEE Transactions on Antennas and Propagation, Vol. 42, No. 10.
6. J. I. Park, K.T. Kim, “A Comparative study on ISAR imaging algorithms for radar target identification”, Progress in Electromagnetic Research, Vol. 108, 155-175, 2010.
7. D. Gracobello, M. G. Christensen, M.N. Murthi, S.H. Jensen, M. Moonene, “Sparse Linear prediction and its applications to speech processing”, IEEE Trans on Audio Speech and LanguageProcessing, Vol. 20, No.5, 2012 .
8. M. Grant, S. Boyd, CVX: Matlab software for Disciplined Convex Programming (web page and software) 2008 [online]. Available : http//Stanford.edu/boyd/cvx.

9. P. C. Hansen, D.P. O'leary, “The use of the L-curve in the regularization of discrete ill-posed problems”, SIAM on Sci. Comp., Vol. 14, No. 6, 1478-1503, 1993.
10. V. C. Chen, H. Ling., Time-frequency transforms for radar imaging and signal analysis, Boston: Artech House, 2002.
11. Kim,K.T., Seo, D. K., and Kim, H. T., (2005). Efficient classification of ISAR images, IEEE Trans. Antennas Propag. ,53, 1611-1621
12. Moving and Stationary Target Acquisition and Recogition (MSTAR) Public Dataset website: https://www.sdms.afrl.af.mil/datasets/mstar/
13. Ganggang Dong Na Wang Gangyao Kuang Yinfa Zhang, Kernel linear representation: application to target recognition in synthetic aperture radar images Journal of Applied Remote Sensing, Vol.8,No.1,2014