Sparsity-Driven Reconstruction Technique for Microwave/Millimeter-Wave Computational Imaging

Lowering latency and processing burden in computational imaging through dimensionality reduction of the sensing matrix

Recent demonstrations have shown that frequency-diverse computational imaging systems can greatly simplify conventional architectures developed for imaging by transferring constraints into the digital layer. Here, in order to limit the latency and processing burden involved in image reconstruction, we propose to truncate insignificant principal components of the sensing matrix that links the measurements to the scene to be imaged. In contrast to recent work using principle component analysis to synthesize scene illuminations, our generic approach is fully unsupervised and is applied directly to the sensing matrix. We impose no restrictions on the type of imageable scene, no training data is required, and no actively reconfigurable radiating apertures are employed. This paper paves the way to the constitution of a new degree of freedom in image reconstructions, allowing one to place the performance emphasis either on image quality or latency and computational burden. The application of such relaxations will be essential for widespread deployment of computational microwave and millimeter wave imagers in scenarios such as security screening. We show in this specific context that it is possible to reduce both the processing time and memory consumption with a minor impact on the quality of the reconstructed images.

A Microwave Three-Dimensional Imaging Method Based on Optimal Wave Spectrum Reconstruction

Limited by the Shannon-Nyquist sampling law, the number of antenna elements and echo signal data of the traditional microwave three-dimensional (3D) imaging system are extremely high. Compressed sensing imaging methods based on sparse representation of target scene can reduce the data sampling rate, but the dictionary matrix of these methods takes a lot of memory, and the imaging has poor quality for continuously distributed targets. For the above problems, a microwave 3D imaging method based on optimal wave spectrum reconstruction and optimization with target reflectance gradient is proposed in this paper. Based on the analysis of the spatial distribution characteristics of the target echo in the frequency domain, this method constructs an orthogonal projection reconstruction model for the wavefront to realize the optimal reconstruction of the target wave spectrum. Then, the inverse Fourier transform of the optimal target wave spectrum is optimized according to the law of the target reflectance gradient distribution. The proposed method has the advantages of less memory space and less computation time. What is more, the method has a better imaging quality for the continuously distributed target. The computer simulation experiment and microwave anechoic chamber measurement experiment verify the effectiveness of the proposed method.

Terahertz Frequency-Scaled Differential Imaging for Sub-6 GHz Vehicular Antenna Signature Analysis

The next generation of connected and autonomous vehicles will be equipped with high numbers of antennas operating in a wide frequency range for communications and environment sensing. The study of 3D spatial angular responses and the radiation patterns modified by vehicular structure will allow for better integration of the associated communication and sensing antennas. The use of near-field monostatic focusing, applied with frequency-dimension scale translation and differential imaging, offers a novel imaging application. The objective of this paper is to theoretically and experimentally study the method of obtaining currents produced by an antenna radiating on top of a vehicular platform using differential imaging. The experimental part of the study focuses on measuring a scaled target using an imaging system operating in a terahertz band—from 220 to 330 GHz—that matches a 5G frequency band according to frequency-dimension scale translation. The results show that the induced currents are properly estimated using this methodology, and that the influence of the bandwidth is assessed.

Linear and Circular UWB Millimeter-Wave and Terahertz Monostatic Near-Field Synthetic Aperture Imaging

Millimeter-wave and terahertz frequencies offer unique characteristics to simultaneously obtain good spatial resolution and penetrability. In this paper, a robust near-field monostatic focusing technique is presented and successfully applied for the internal imaging of different penetrable geometries. These geometries and environments are related to the growing need to furnish new vehicles with radar-sensing devices that can visualize their surroundings in a clear and robust way. Sub-millimeter-wave radar sensing offers enhanced capabilities in providing information with a high level of accuracy and quality, even under adverse weather conditions. The aim of this paper was to research the capability of this radar system for imaging purposes from an analytical and experimental point of view. Two sets of measurements, using reference targets, were performed in the W band at 100 GHz (75 to 110 GHz) and terahertz band at 300 GHz (220 to 330 GHz). The results show spatial resolutions of millimeters in both the range (longitudinal) and the cross-range (transversal) dimensions for the two different imaging geometries in terms of the location of the transmitter and receiver (frontal or lateral views). The imaging quality in terms of spatial accuracy and target material parameter was investigated and optimized.