*4.2. Experimental Results*

In order to further verify the effectiveness of the proposed SSCF technique, equivalent verification experiments in the laboratory environment were carried out. The schematic diagram of the THz radar system and the photograph of the front-end setup is shown in Figure 10. The five antennas were arranged in two rows, with three receiving antennas in the upper row and one transmitting antenna and one receiving antenna in the other row. R1 and R2 formed the vertical interferometric baseline, and R2 and R3 formed the horizontal interferometric baseline. Both the vertical and horizontal baseline lengths were 2.1 cm. R4 was ignored in the InISAR experiment. Besides InISAR imaging, this system also had many other potential applications such as InSAR imaging, ViSAR imaging, and micromotion target 3-D imaging. The THz radar system was based on the linear frequency modulated pulse principle. A chirped signal ranging from 217.1 to 222.1 GHz was transmitted, and the echo signals were received by the four receiving antennas.

**Figure 10.** The terahertz radar system. (**a**) Schematic diagram. (**b**) Photograph of the front-end setup.

The experimental configuration is shown in Figure 11. In the experiment, the THz radar was put on a one-dimensional horizontal guide platform, and the velocity of radar was 1 m/s. The vertical distance between the radar and the target was 10 m, and the initial connection from the target to the radar platform was perpendicular to the motion direction. This was a typical SAR imaging scenario, but it was equivalent in that the radar was static, and the target moved along the horizontal direction, which matched the InISAR scenario described in this paper. To decrease the SNR of the radar echoes, the power transmitted from the radar was reduced to 10 mW. The pulse width was 163.8 μs, the pulse repetition frequency was 2500 Hz, the sampling frequency was 12.5 MHz, and the data acquisition time was 1 s. The target was an Airbus A380 model. The length, wingspan, and height of the model were 45, 52, and 17 cm, respectively.

**Figure 11.** The experimental configuration.

The reflected signals received by R1, R2, and R3 were used to form the InISAR images, and the reference signal was the reflected signal of a corn reflector located at the same position of the airplane model received by R2. Thus, the imaging geometry is the same as the discussed L-shaped three-antenna configuration. During the imaging process, nonlinearity of the signal frequency and the inconsistencies of the amplitudes and phases among channels were compensated together with the reference signal, and a phase gradient autofocus algorithm [36] was adopted to compensate the influence of guide platform vibration. The ISAR images of channels R2 and R3 were interpolated three times and shown in Figure 12. Taking the strong scattering center at the right wing as an example, there are five Doppler cells that deviated along the azimuth direction.

**Figure 12.** The ISAR images of A380 model. (**a**) Channel R2. (**b**) Channel R3.

Based on a threshold of 3 dB, the ISAR image of six SSCs were extracted, as indicated in Figure 12. The spatial spectrums of these SSCs were then acquired. The phase difference curves of spatial spectrums between channel R2 and R3 are shown in Figure 13, with (a) and (b) corresponding to the conditions before and after phase unwrapping processing, respectively. As illustrated in Figure 13a, all curves were linear, and the phase wrapping position of each curve was constant. With our experimental parameters, the maximum allowed error in velocity estimation for precise image registration was 0.0812 m/s on the basis of Equation (22). Based on the motion measurement curves in Figure 13b, velocity along the horizontal direction was estimated as 0.9737 m/s, which satisfied the estimated

precision of parameters for image registration. After compensating the velocity to the echo signal of channel R3, image registration was achieved. Finally, the InISAR imaging results were obtained, as shown in Figure 14, with (a), (b), (c), and (d) corresponding to the 3-D view and projections on the *xoy*, *xoz*, and *yoz* planes, respectively. From the InISAR imaging results, we saw that the key parts in the A380 model such as the engine, wing, and vertical fin could be clearly identified, and the imaging results were clear and close to the real airplane model. These results further verified the effectiveness of the proposed SSCF method. In this chapter, we did not compare the performance between our method and the conventional method because the baseline length in this equivalent verification experiment could not be any shorter. Phase wrapping in the motion measurement curves was inevitable. In this condition, the conventional method was not applicable.

**Figure 13.** Phase difference curves of the spatial spectrum. (**a**) Before phase unwrapping. (**b**) After phase unwrapping.

**Figure 14.** InISAR imaging results of the airplane model. (**a**) 3-D view; (**b**) *xoy* view projection; (**c**) *xoz* view projection; and (**d**) *yoz* view projection.
