(a) | output: s | output: t | ||
input: s | $ 1 - r_s $ | $ r_s $ | ||
input: t | $ r_t $ | $ 1 - r_t $ | ||
(b) | output: s | output: t | uncertain | |
input: s | $ 1 - r'_s -r''_s $ | $ r'_s $ | $ r''_s $ | |
input: t | $ r'_t $ | $ 1 - r'_t- r''_t $ | $ r''_t $ |
Distinguishing between the instantaneous and delayed scatterers in synthetic aperture radar (SAR) images is important for target identification and characterization. To perform this task, one can use the autocorrelation analysis of coordinate-delay images. However, due to the range-delay ambiguity the difference in the correlation properties between the instantaneous and delayed targets may be small. Moreover, the reliability of discrimination is affected by speckle, which is ubiquitous in SAR images, and requires statistical treatment.
Previously, we have developed a maximum likelihood based approach for discriminating between the instantaneous and delayed targets in SAR images. To test it, we employed simple statistical models. They allowed us to simulate ensembles of images that depend on various parameters, including aperture width and target contrast.
In the current paper, we enhance our previously developed methodology by establishing confidence levels for the discrimination between the instantaneous and delayed scatterers. Our procedure takes into account the difference in thresholds for different target contrasts without making any assumptions about the statistics of those contrasts.
Citation: |
Figure 1. Plots of $ \left\langle{|I_t|^2}\right\rangle $ and $ \left\langle{|I_s|^2}\right\rangle $ for different values of $ \zeta_{\max} $ and $ \kappa $, see (39) and (27). The dashed lines passing through the origin indicate the ambiguity direction, see (14), (18). For the middle row of plots, the condition $ \kappa \zeta_{\max} \gtrsim 20 $ (see (40)) is satisfied, and the difference in the orientation of the parallelogram-shaped level lines is more apparent than for the top and bottom rows
Figure 2. Simulated coordinate-delay SAR images with different contrasts. We use $ \kappa = 2.5 $ and $ \zeta_{\max} = 5\pi $, which corresponds to the middle row in Fig. 1. To build each image, we sample $ \zeta_\text{d} $ and $ \psi_\text{d} $ with a step of $ \pi $ each, and for every $ \zeta_\text{d} $ (i.e., each ambiguity line) generate the multivariate circular Gaussian image components according to (36) with the weights $ \sigma^2_ \alpha K_ \alpha $ calculated via (43). The total images are then computed according to the appropriate expression in (42). Simulation of multivariate normal random variables is performed using the $ \text{MATLAB}^© $ function $\mathtt{mvnpdf}$
Figure 3. Cumulative distribution functions (cdf) for ensembles generated from the s-model and t-model. (a) Examples of cdfs for $ \log \breve p_s $ and $ \log \breve p_t $, see (54). The notations $ \text{cdf}_s[\ldots] $ and $ \text{cdf}_t[\ldots] $ are similar to those in (60). (b) Examples of cdfs for $ l = \log \breve p_t - \log \breve p_s $, see (55) and (60). According to (61) (see also Table 1(a)), we have $ r_t = \text{cdf}_t(0; q) $ and $ r_s = 1- \text{cdf}_s(0; q) $. (c) A zoom-in to the central part of panel (b). The entries $ r'_s $, $ r''_s $, $ r'_t $, and $ r''_t $ from Table 1(b) are determined according to (63) and (65) with $ p = 0.05 $
Figure 4. Separation between the graphs of $ \text{cdf}_s(\cdot\, ;q) $ and $ \text{cdf}_t(\cdot\, ;q) $ for different values of $ \kappa $ and $ q $. The thick colored vertical bars indicate the percentage of uncertain classifications for the ensembles generated from the s-model (the left set of bars in each plot) and t-model (the right set of bars), see also (65) and Fig. 3(c)
Figure 5. Dependence of the discrimination quality on $ \zeta_{\max} $, see (39), for two different target contrasts. The left column of plots corresponds to algorithm (56) and Table 1(a), and the right column corresponds to algorithm (57) and Table 1(b). The dashed vertical lines are drawn at $ \kappa \zeta_{\max} = b_\Phi \approx 23 $, cf. (40). The lower and upper parts of the colored panels represent ensembles generated from the s-model and t-model, respectively. The percentage of correct classifications is shown in green (two different shades are used to distinguish between the ensemble types), incorrect in red, and uncertain in yellow (only the right column of plots)
Figure 6. Dependence of discrimination quality on $ \kappa $, see (27). The notations are the same as in Fig. 5
Table 1.
Confusion matrices: (a) for classification using algorithm (56); (b) for classification using algorithm (57). The rows correspond to the models in (42), whereas the columns indicate the outcomes of a particular classification algorithm. The entries are relative frequencies of the events calculated for two ensembles with the same contrast, i.e.,
(a) | output: s | output: t | ||
input: s | $ 1 - r_s $ | $ r_s $ | ||
input: t | $ r_t $ | $ 1 - r_t $ | ||
(b) | output: s | output: t | uncertain | |
input: s | $ 1 - r'_s -r''_s $ | $ r'_s $ | $ r''_s $ | |
input: t | $ r'_t $ | $ 1 - r'_t- r''_t $ | $ r''_t $ |
[1] | J. C. Allen and S. L. Hobbs, Spectral estimation of non-stationary white noise, J. Franklin Inst. B, 334 (1997), 99-116. doi: 10.1016/S0016-0032(96)00060-9. |
[2] | M. Basu, Gaussian-based edge-detection methods–a survey, IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews), 32 (2002), 252-260. doi: 10.1109/TSMCC.2002.804448. |
[3] | C. M. Bishop, Pattern Recognition and Machine Learning, Information Science and Statistics, Springer, New York, 2006. |
[4] | B. Borden, Dispersive scattering for radar-based target classification and duct-induced image artifact mitigation, NATO Symposium on Non-Cooperative Air Target Identfication Using Radar, Mannheim, Germany, 1998, 14.1–14.7. |
[5] | J. Canny, A computational approach to edge detection, IEEE Transactions on Pattern Analysis and Machine Intelligence, PAMI-8 (1986), 679-698. doi: 10.1109/TPAMI.1986.4767851. |
[6] | S. Chen and H. Wang, SAR target recognition based on deep learning, 2014 International Conference on Data Science and Advanced Analytics (DSAA), IEEE, Shanghai, China, 2014, 541–547. doi: 10.1109/DSAA.2014.7058124. |
[7] | V. C. Chen and H. Ling, Time-Frequency Transforms for Radar Imaging and Signal Analysis, Artech House Radar Library, Artech House, Norwood, MA, 2002. |
[8] | M. Cheney and B. Borden, Fundamentals of Radar Imaging, CBMS-NSF Regional Conference Series in Applied Mathematics, Vol. 79, Society for Industrial and Applied Mathematics (SIAM), Philadelphia, PA, 2009. doi: 10.1137/1.9780898719291. |
[9] | F. W. J. Olver, A. B. Olde Daalhuis, D. W. Lozier, B. I. Schneider, R. F. Boisvert, C. W. Clark, B. R. Miller, B. V. Saunders, H. S. Cohl, and M. A. McClain, (eds.). NIST Digital Library of Mathematical Functions, http://dlmf.nist.gov/, Release 1.0.24 of 2019-09-15. |
[10] | M. Ferrara, A. Homan and M. Cheney, Hyperspectral SAR, IEEE Trans. Geoscience and Remote Sensing, 55 (2017), 1682–1695. doi: 10.1109/TGRS.2016.2629265. |
[11] | J. Garnier and K. Sølna, A multiscale approach to synthetic aperture radar in dispersive random media, Inverse Problems, 29 (2013), 054006, (18pp). doi: 10.1088/0266-5611/29/5/054006. |
[12] | M. Gilman and S. Tsynkov, Mathematical analysis of SAR imaging through a turbulent ionosphere, Application of Mathematics in Technical and Natural Sciences: 9th International Conference for Promoting the Application of Mathematics in Technical and Natural Sciences– AMiTaNS'17, Vol. 1895, AIP Conference Proceedings, Albena, Bulgaria, 2017, 020003 (23pp). doi: 10.1063/1.5007357. |
[13] | M. Gilman, E. Smith and S. Tsynkov, Transionospheric Synthetic Aperture Imaging, Applied and Numerical Harmonic Analysis, Birkhäuser/Springer, Cham, Switzerland, 2017. doi: 10.1007/978-3-319-52127-5. |
[14] | M. Gilman and S. Tsynkov, Detection of delayed target response in SAR, Inverse Problems, 35 (2019), 085005, (38pp). doi: 10.1088/1361-6420/ab1c80. |
[15] | J. W. Goodman, Statistical properties of laser speckle patterns, Topics in Applied Physics: Laser Speckle and Related Phenomena (ed. J. C. Dainty, ), Vol. 9, Springer-Verlag, Berlin, 1984, 9–75. |
[16] | T. Hastie, R. Tibshirani and J. Friedman, The Elements Of Statistical Learning. Data Mining, Inference, and Prediction, 2$^nd$ edition, Springer Series in Statistics, Springer, New York, 2009. doi: 10.1007/978-0-387-84858-7. |
[17] | A. Krizhevsky, I. Sutskever and G. E. Hinton, ImageNet classification with deep convolutional neural networks, Advances in Neural Information Processing Systems 25, Curran Associates, Inc., 2012, 1097–1105. |
[18] | D. Marr and E. Hildreth, Theory of edge detection, Proc. R. Soc. Lond. B, 207 (1980), 187-217. doi: 10.1098/rspb.1980.0020. |
[19] | W. Mendenhall and R. L. Scheaffer, Mathematical Statistics with Applications, Duxbury Press, North Scituate, Mass., 1973. |
[20] | F. J. Meyer, K. Chotoo, S. D. Chotoo, B. D. Huxtable and C. S. Carrano, The influence of equatorial scintillation on L-band SAR image quality and phase, IEEE Trans. Geoscience and Remote Sensing, 54 (2016), 869-880. doi: 10.1109/TGRS.2015.2468573. |
[21] | D. Mishkin, N. Sergievskiy and J. Matas, Systematic evaluation of convolution neural network advances on the Imagenet, Computer Vision and Image Understanding, 161 (2017), 11-19. doi: 10.1016/j.cviu.2017.05.007. |
[22] | A. S. Monin and A. M. Yaglom, Statistical Fluid Mechanics: Mechanics of Turbulence, Dover Publications, Inc., Mineola, NY, 2007. |
[23] | C. Oliver and S. Quegan, Understanding Synthetic Aperture Radar Images, Artech House Remote Sensing Library, Artech House, Boston, 1988. |
[24] | S. M. Rytov, Y. A. Kravtsov and V. I. Tatarskii, Principles of Statistical Radiophysics, Volume 4. Wave Propagation Through Random Media, Springer-Verlag, Berlin, 1989. |
[25] | T. Scarnati and B. Lewis, A deep learning approach to the synthetic and measured paired and labeled experiment (SAMPLE) challenge problem, Algorithms for Synthetic Aperture Radar Imagery XXVI, 10987 (2019), 109870G. doi: 10.1117/12.2523458. |
[26] | F. T. Ulaby and M. C. Dobson, Handbook of Radar Scattering Statistics for Terrain, Artech House Remote Sensing Library, Artech House, Norwood, MA, USA, 1989. |
[27] | X. X. Zhu, D. Tuia, L. Mou, G.-S. Xia, L. Zhang, F. Xu and F. Fraundorfer, Deep learning in remote sensing: A comprehensive review and list of resources, IEEE Geoscience and Remote Sensing Magazine, 5 (2017), 8-36. doi: 10.1109/MGRS.2017.2762307. |
[28] | D. Ziou and S. Tabbone, et al., Edge detection techniques-an overview, Pattern Recognition and Image Analysis C/C of Raspoznavaniye Obrazov I Analiz Izobrazhenii, 8 (1998), 537-559. |
Plots of
Simulated coordinate-delay SAR images with different contrasts. We use
Cumulative distribution functions (cdf) for ensembles generated from the s-model and t-model. (a) Examples of cdfs for
Separation between the graphs of
Dependence of the discrimination quality on
Dependence of discrimination quality on