Real Data Processing Results
In this section, the real Ku-band airborne SAR data is used to verify the effectiveness of the proposed strategy. This scene includes roads in both the cross- and along-track directions and a crossroad. Seven moving targets with different motion parameters, which belong to Type I, II, and III, are found in the scene. Moreover, the velocity errors of both the platform and the moving target exist in the data, and they must be compensated during the imaging.
The seven moving targets, labeled from T1 to T7, are illustrated in both the range-Doppler domain and the image domain in Fig. 3.8. The blue lines in Fig. 3.8a indicate the spectrum of the clutter. According to Fig. 3.8a, T1, T2, T3, and T4 move along the cross-track direction. The trajectories of T1 and T2 are located at the high-band of the PRF, and T3 is partially submerged by the clutter in the range-Doppler domain. The trajectory of T4 can be distinguished from the clutter in Fig. 3.8a due to its different slope, whereas it is completely submerged by the clutter. The detection of T4 is the most difficult problem in a single-antenna SAR system. T1 and T2 cannot be seen in Fig. 3.8b since their energy is weaker compared with that of the clutter. T5, T6 and T7 have only along-track velocities and can be found defocused in the azimuth direction in Fig. 3.8b, whereas they cannot be detected in the range-Doppler domain due to that they have no cross-track velocities. We can conclude from Fig. 3.8 that T1 and T2 belong to Type I, T3 belong to Type II, and T4, T5, T6, and T7 belong to Type III. The detection of the moving targets from different types must be conducted differently.
The CFAR detection result of T1 and T2 is shown in Fig. 3.9. The spectrum of the clutter is suppressed by a high-pass filter, and then the rest echo is compressed in the azimuth direction. Particularly, the image of T2 splits into two parts since the spectrum of T2 exceeds the limit of the PRF. It can be noted that moving target of Type I can be successfully indicated by the two-step GMTI algorithm in the range-Doppler domain.
Fig. 3.8 Illustrations of the moving targets in the scene. a The moving targets in the range- Doppler domain. b The moving targets in the stationary image
Fig. 3.9 CFAR detection result of T1 and T2
After the detection of T1 and T2, their energy is extracted, and the remaining signal is separated into two sub-apertures. The sub-aperture imaging results are shown in Fig. 3.10. It can be noted from Fig. 3.10 that the locations of the moving targets are different in the two sub-images. The moving targets can be detected by
Fig. 3.10 Results of the sub-aperture imaging. a Imaging of T3 and T4 in Sub-image I. b Imaging of T3 and T4 in Sub-image II. c Imaging of T5, T6, and T7 in Sub-image I. d Imaging of T5, T6, and T7 in Sub-image II
the complex image subtraction. T3 is not found in Fig. 3.10a because its energy locates only in half of the PRF, as shown in Fig. 3.8a.
The result of the moving target spectrum extraction is shown in Fig. 3.11. T3 and T4 are used as an example to show the effectiveness of the proposed algorithm because the extraction of targets that are submerged by the clutter is very difficult. It can be noted from Fig. 3.11a that the trajectories of T3 and T4 are extracted by the proposed algorithm. Although the energy of the clutter is not completely eliminated, the SCR after the extraction is clearly improved. The imaging of the clutter after the extraction is shown in Fig. 3.11b. The smeared images of the moving targets are disappeared, and the stationary area that masked by the moving targets is reverted.
Fig. 3.11 Results of the moving target extraction. a The extracted trajectory of T3 and T4 in the range-Doppler domain. b Partial imaging of the scene without moving targets T3 and T4