Active mmWave imaging is used to scan personnel in order to detect dangerous items hidden under clothing in airports, customs facilities and other key places. To obtain clear, high resolution images, an algorithm is first verified through analysis, then a set of requirements for core system hardware is derived from the holographic algorithm model. Finally, a high-performance mmWave three-dimensional imaging system is developed and high-quality images are obtained. The resolution of Ka-band images is as high as 5 mm.
The mmWave spectrum, 30 to 300 GHz frequency, can penetrate through clothing, packaging and leather.1 This makes it a potentially a viable means for the detection of hidden items such as weapons, explosives, drugs and other contraband.2 Wavelengths at these frequencies are relatively short (1 to 10 mm), providing high spatial resolution. Radiation at these frequencies is also non-ionizing, causing no harm to the human body at appropriate power levels. Thus, mmWave imaging systems are well suited for use in human security screening.3-8
Existing mmWave imaging systems are of two fundamental types: active and passive. Passive systems use compact structures to identify temperature contrast. These are capable of real-time 2D imaging, but cannot reliably distinguish concealed items from the human body in an indoor environment and are unable to generate three-dimensional (3D) images.
Active mmWave human security systems can detect various dangerous items hidden under human clothing by analyzing the differences in electromagnetic energy scattered from objects in time and frequency, spatial, polarization and power domains. Moreover, active systems enable the formation of fully focused 3D images from data gathered optically or via a Fourier transform over a 2D aperture, with superior resolution.
Much research has been reported on mmWave imaging systems and imaging methods,9-12 with varying levels of performance. However, from the perspective of an actual application of hardware processing, research on image resolution optimization by changing hardware parameters has been rarely discussed.13-18
This article describes a mmWave (27 to 32 GHz) 3D imaging system for human safety inspection. It begins with a discussion of the algorithmic model of the system, which is the basis for the analysis of key system parameters. Image resolution and image quality are improved by the optimization of parameters, such as the distribution error of the antenna array element and the linearity of the signal emitter. Next, the system link analysis and radio frequency (RF) transceiver module design are introduced. Based on the modeling and analysis, a high-performance MMW 3D imaging system is built to obtain high-quality mmWave images. Resolution of the Ka-band images is 5 mm.
TARGET ECHO SIGNAL MODEL
Figure 1 shows a 3D rectangular coordinate system for an antenna continuously scanning in the vertical and horizontal directions, synthesizing the 2D scanning plane, obtaining high-quality images and obtaining the target depth of field. Let R1 be the distance from the transmitting antenna to the target point P and R2 be the distance from the receiving antenna to the target point P. In this configuration, the time delay between the target echo signal received by the receiving antenna and the transmitted signal can be expressed by:
![](/ext/resources/images/Figures/2023/Dec/OL/m1.jpg)
![Figure 1](/ext/resources/images/Figures/2023/Dec/OL/f1.jpg)
Figure 1 Coordinate system for 3D imaging.
If the emitted signal is a linear frequency modulated (LFM) signal, then σ (x,y,z) is the scattering coefficient of the target. If attenuation of the narrowband signal in the near field is ignored, then the amplitude of the received signal collected by the monostatic radar is determined by the scattering coefficient of the target point, where the phase is determined by the round-trip path from this point to the antenna array element.10, 11 After the time delay from the target scattering point and the frequency mixing of the coherent signal, the fundamental frequency echo is given by:
Where:
Δ0 is the difference between the reference delay and the actual delay and Δ02 is the residual video phase, which can be removed by Dskew filtering in the frequency domain (see Figure 2).
![Figure 2](/ext/resources/images/Figures/2023/Dec/OL/f2.jpg)
Figure 2 Algorithm flow for 3D reconstruction.
The baseband echo is a single frequency determined by the frequency modulation slope and echo delay. Correspondingly, Fourier transforms and phase compensation are performed in the vertical and horizontal dimensions, respectively. Finally, through interpolation and 3D inverse Fourier processing, a 3D image of the region is obtained.
DERIVATION OF KEY SYSTEM PARAMETERS
The mmWave human body detection system is an imaging device, so imaging resolution determines its detection ability. To obtain clear high resolution images, the algorithm is first verified, then the achievable parameter range of the hardware is derived from a holographic algorithm model.
The specific method is as follows: 1) add hardware error parameters to the 3D model to obtain the image resolution for different parameter values and 2) derive the required values of key parameters for the antenna array, RF transceiver and scanning machine frame by analyzing the peak-to-sidelobe ratio (PSLR), integrated sidelobe ratio (ISLR) and 3 dB image width.
Too high a demand on hardware performance increases hardware design difficulty and processing costs. At the same time, it is also necessary to prevent the deterioration of images due to overly relaxed requirements. The focus of this article is to establish requirements for hardware design through algorithms by determining a reasonable range of performance that minimizes design difficulty and reduces design cost while providing high image resolution. To that end, it looks at minimizing the error sources shown in Figure 3.
![Figure 3](/ext/resources/images/Figures/2023/Dec/OL/f3.jpg)
Figure 3 Imaging radar system error sources.
ANTENNA ARRAY ERRORS
Element design refers to the design of a single transceiver antenna. In Figure 4, Tn and Rn are individual antenna elements, and multiple antenna elements are arranged linearly to form an antenna array. Antenna polarization is a parameter that describes the spatial direction of the electromagnetic wave vector radiated by the antenna. The transmitting and receiving antennas of this system are all horizontally polarized. The array channels are calibrated in amplitude and phase.
One vertical column is the transmitting antenna array and the other column is the receiving antenna array. When one antenna element transmits, two adjacent elements receive. It is assumed that signal transmission and reception are at the midpoint of the connecting line between the transmitting and receiving elements, which is equivalent to a virtual sampling point between each transmitting and receiving antenna element. Therefore, the interval dy between effective sampling points is half of the interval between single row antenna units, which is 5 mm. In this way, only 192 transmitting elements and 193 receiving elements are needed to obtain 384 data samples. Not only is the number of antenna units reduced by half, but also the spacing between antenna units is doubled. This reduces processing and increases isolation between antenna elements.