Multi-view TWRI scene reconstruction using a joint Bayesian sparse approximation model

With recent advances in both algorithm and component technologies, through-the-wall radar imaging (TWRI) is emerging as an affordable sensor technology in civil and military settings. Design of appropriate waveforms is important for improving the performance of TWRI systems. We present optimal waveforms, based on the matched illumination concept, for detection of targets inside enclosed structures using a monostatic multi-antenna radar system. We evaluate the effect of these optimal detection waveforms on the performance of through-the-wall radar imaging. The design flexibility for transmit signal duration and its role in the ultimate range resolution of the TWRI system is investigated. Supporting examples based on numerical electromagnetic modeling are also provided.

Through-the-wall radar imaging (TWRI) provides the capability to see inside buildings using electromagnetic waves for both defence and civilian applications. Unlike traditional free-space radar operation, the presence of walls introduces unique challenges that need to be addressed to render TWRI a viable technology. In this chapter, we present a review of various approaches for building layout determination and imaging of stationary indoor scenes. These approaches effectively address the associated challenges to provide effective and reliable radar operation. Specifically, we consider both ground-based and airborne radar operation and discuss both non-adaptive time-domain and frequency-domain beamforming approaches for through-the-wall image formation of stationary scenes and feature extraction-based building layout determination methods.

To achieve high resolution through-the-wall radar imaging (TWRI), long wideband antenna arrays need to be considered, thus resulting in massive amounts of data. Compressive sensing (CS) techniques resolve this issue by allowing image reconstruction using much fewer measurements. The performance of different CS algorithms, when applied to TWRI, has not been investigated in a comprehensive and comparative manner. In this paper, popular CS algorithms are evaluated, to see which are most suitable for TWRI applications. As for the evaluation criteria, the notion of F1-score is adopted and used in the context of TWRI; thus emphasising the algorithms ability to reconstruct an image with correctly detected targets. Algorithms responses to different levels of signal-to-noise ratio (SNR) and compression rate are evaluated. Numerical results show that for systems with low SNR, alternating direction based algorithms work better than others. When the SNR is high, algorithms depending on spectral gradient-projection methods give good results even with high compression rates.

J-vector has been proved to be very effective in text-dependent speaker verification with short-duration speech. However, the current state-of-the-art back-end classifiers, e.g. joint Bayesian model, cannot make full use of such deep features. In this paper, we generalize the standard joint Bayesian approach to model the multi-faceted information in the j-vector explicitly and jointly. In our generalization, the j-vector was modeled as a result derived by a generative Double Joint Bayesian (DoJoBa) model, which contains several kinds of latent variables. With DoJoBa, we are able to explicitly build a model that can combine multiple heterogeneous information from the j-vectors. In verification step, we calculated the likelihood to describe whether the two j-vectors having consistent labels or not. On the public RSR2015 data corpus, the experimental results showed that our approach can achieve 0.02\% EER and 0.02\% EER for impostor wrong and impostor correct cases respectively.

Through-the-wall radar imaging (TWRI) is usually associated with the use of ultra wideband (UWB) signals to allow for wall penetration and high resolution. It is difficult to keep the antenna pattern fixed at all frequencies. Due to the large bandwidth and associated big data, compressive sensing (CS) is used for TWRI. We propose to embed the directivity of the antenna in the designed sparsity dictionary. The results show improved image reconstruction.

In this paper, we provide a review of the different approaches used to detect activities inside a building using through-the-wall radar. We classify two main types of radar that are used in detection activities behind the wall. Stepped frequency and pulse compression radar are the most frequently used types of radar in through-the-wall radar imaging (TWRI). In addition, a preference has been shown for using frequencies in the range of 1–3 GHz. Though a high volume of research is being done in this area, there are still many open issues. Detectability of the human being, wall modeling, and target differentiation have been identified as the main open issues in the field of TWRI.

Polarimetric through-the-wall radar imaging (TWRI) system has the enhancing performance in the detection, imaging, and classification of concealed targets behind the wall. We propose a group sparse basis pursuit denoising- (BPDN-) based imaging approach for polarimetric TWRI system in this paper. The proposed imaging method combines the spectral projection gradient L1-norm (SPGL1) algorithm with the nonuniform fast Fourier transform (NUFFT) technique to implement the imaging reconstruction of observed scene. The experimental results have demonstrated that compared to the existing compressive sensing- (CS-) based imaging algorithms, the proposed NUFFT-based SPGL1 algorithm can significantly reduce the required computer memory and achieve the improved imaging reconstruction performance with the high computational efficiency.

Through-the-wall radar imaging (TWRI) is currently an active and growing area of research and development because the detection and localization of targets in enclosed structures has many civilian and military applications such as rescue operations in case of fire or after earthquakes. However, contrary to free-space imaging, methods needed for TWRI are not a straight-forward process and face many challenges, all imposed by the presence of the building walls [1]. One problem pertains to frequency dependent attenuation and phase dispersion caused by exterior walls. This limits the useful frequency range to the UHF band with its lower frequency limited by the antenna size. Consequently, the bandwidth and image resolution of TWR systems is limited. The second issue is related to the scattering from the building walls and the multipath signals they create. The specular reflection and multiple bounces within the wall create a very strong signal much larger than that of the target behind the wall which normally has a small radar cross section (e.g. human). This leads to two problems: If the target is close to the wall, it's return can be completely covered by the wall return or if the target is further away from the wall, it can still be covered by the strong multiple bounce response or the point spread function sidelobes of the wall return. It is possible to reduce the sidelobe level but at the expense of range resolution by windowing the frequency response of the radar return.

A three-dimensional (3D) beamformer for through-the-wall radar imaging (TWI) for a general multilayered wall case is presented. The far field layered medium Green's function is incorporated in the 3D TWI algorithm for the compensation of the wall effect. Numerical results show that the 3D beamformer can provide high quality focused image of the target behind the walls.

Through-the-wall radar imaging (TWRI) applications allow accurate target localization and high-resolution imaging. However, multipath propagation generates challenges to the image reconstruction procedure. With distortions in the received radar signals, traditional imaging algorithms are not able to acquire high-resolution images. The unpredictability of the indoor scattering environment makes this even worse. In this paper, a novel block orthogonal matching pursuit (BOMP)-based group-sparsity reconstruction algorithm combined with particle swarm optimization (PSO) is proposed for reliable scene reconstruction. The proposed imaging algorithm can recover the image of targets by exploiting multipath propagation and simultaneously estimating wall parameters with high accuracy. The effectiveness of the proposed imaging method has been further demonstrated via simulation results.

Bayesian sparse hierarchical model for image denoising

In this paper, measured results for complex permittivity of some commonly used building walls under difierent hydration (wetness) levels are presented and a simple hybrid measurement and electromagnetic modeling approach for the estimation of power returns from targets located behind walls in various through-the-wall radar imaging (TWRI) scenarios is discussed. The radar cross section (RCS) of some typical targets of interest, such as an AK47 assault ri∞e and human, are flrst investigated in free-space using numerical electromagnetic modeling. A modifled radar range equation, which analytically accounts for the wall efiects, including multiple re∞ections within a given homogeneous or layered wall, is then employed in conjunction with wideband measured parameters of various common wall types, to estimate the received power versus frequency from modeled targets of interest. The proposed technique, which can be helpful in TWRI dynamic-range system design consideration, is, in principle, applicable to both bistatic and monostatic operations. The results for various wall types, including drywall, brick, solid concrete, and cinder block, the latter two under both wet and dry conditions, are presented.

Through-the-wall radar imaging (TWRI) enables critical tasks such as search and rescue, surveillance, and health monitoring. However, strong wall reflections often obscure weak human signatures. This work presents a simulation-driven framework for detecting stationary and breathing individuals at low frequencies behind walls. Synthetic radar datasets were generated in MATLAB, modeling multiple wall types and clutter conditions. A convolutional neural network (CNN)-based U-Net was developed, trained, and validated to suppress wall clutter and enhance target visibility. The model effectively distinguishes human responses from dominant wall echoes, yielding over 30 dB improvement in signal-to-clutter ratio. Tests on synthetic and emulated radar data further confirm its ability to recover weak target echoes while maintaining real-time throughput within FPGA power limits. Results demonstrate the robustness and efficiency of the proposed deep learning approach, establishing a strong foundation for embedded implementation and practical field deployment in complex through-wall scenarios.

BackgroundIt has been shown that gene expression in human tissues is heritable, thus predicting gene expression using only SNPs becomes possible. The prediction of gene expression can offer important implications on the genetic architecture of individual functional associated SNPs and further interpretations of the molecular basis underlying human diseases.MethodsWe compared three types of methods for predicting gene expression using only cis-SNPs, including the polygenic model, i.e. linear mixed model (LMM), two sparse models, i.e. Lasso and elastic net (ENET), and the hybrid of LMM and sparse model, i.e. Bayesian sparse linear mixed model (BSLMM). The three kinds of prediction methods have very different assumptions of underlying genetic architectures. These methods were evaluated using simulations under various scenarios, and were applied to the Geuvadis gene expression data.ResultsThe simulations showed that these four prediction methods (i.e. Lasso, ENET, LMM and BSLMM) behaved best when their respective modeling assumptions were satisfied, but BSLMM had a robust performance across a range of scenarios. According to R2 of these models in the Geuvadis data, the four methods performed quite similarly. We did not observe any clustering or enrichment of predictive genes (defined as genes with R2 ≥ 0.05) across the chromosomes, and also did not see there was any clear relationship between the proportion of the predictive genes and the proportion of genes in each chromosome. However, an interesting finding in the Geuvadis data was that highly predictive genes (e.g. R2 ≥ 0.30) may have sparse genetic architectures since Lasso, ENET and BSLMM outperformed LMM for these genes; and this observation was validated in another gene expression data. We further showed that the predictive genes were enriched in approximately independent LD blocks.ConclusionsGene expression can be predicted with only cis-SNPs using well-developed prediction models and these predictive genes were enriched in some approximately independent LD blocks. The prediction of gene expression can shed some light on the functional interpretation for identified SNPs in GWASs.

ABSTRACTFinite memory sources and variable‐length Markov chains have recently gained popularity in data compression and mining, in particular, for applications in bioinformatics and language modelling. Here, we consider denser data compression and prediction with a family of sparse Bayesian predictive models for Markov chains in finite state spaces. Our approach lumps transition probabilities into classes composed of invariant probabilities, such that the resulting models need not have a hierarchical structure as in context tree‐based approaches. This can lead to a substantially higher rate of data compression, and such non‐hierarchical sparse models can be motivated for instance by data dependence structures existing in the bioinformatics context. We describe a Bayesian inference algorithm for learning sparse Markov models through clustering of transition probabilities. Experiments with DNA sequence and protein data show that our approach is competitive in both prediction and classification when compared with several alternative methods on the basis of variable memory length.