Multi-polarization through-the-wall radar imaging using joint Bayesian compressed sensing

Abstract

This paper presents a new image formation method for multi-polarization through-the-wall radar imaging. The proposed method combines wall clutter mitigation and scene reconstruction in a unified framework using multitask Bayesian compressed sensing. First, the radar signals are jointly recovered using Bayesian compressed sensing in the wavelet domain. Then, a subspace projection method is employed to mitigate the front wall reflections. This is followed by principal component analysis, which is used to compress the remaining wavelet coefficients and remove noise. A linear model is developed which relates the compressed wavelet coefficients directly to the image of the scene. For scene reconstruction, multitask Bayesian compressed sensing is further applied to simultaneously form the images associated with all polarimetric channels. Experimental results based on real radar data demonstrate that the proposed method improves image quality by enhancing target reflections and attenuating background clutter. Abstract-This paper presents a new image formation method for multi-polarization through-the-wall radar imaging. The proposed method combines wall clutter mitigation and scene reconstruction in a unified framework using multitask Bayesian compressed sensing. First, the radar signals are jointly recovered using Bayesian compressed sensing in the wavelet domain. Then, a subspace projection method is employed to mitigate the front wall reflections. This is followed by principal component analysis, which is used to compress the remaining wavelet coefficients and remove noise. A linear model is developed which relates the compressed wavelet coefficients directly to the image of the scene. For scene reconstruction, multitask Bayesian compressed sensing is further applied to simultaneously form the images associated with all polarimetric channels. Experimental results based on real radar data demonstrate that the proposed method improves image quality by enhancing target reflections and attenuating background clutter. I. INTRODUCTIONThrough-the-wall radar imaging (TWRI) is emerging as a viable sensing technology supporting a range of civilian and military applications, such as search-and-rescue, law enforcement, and urban surveillance and reconnaissance. It can be used for various purposes, including determining the building layout, and locating and identifying stationary objects behind walls. In the past years, numerous studies have been conducted in modeling and imaging stationary and moving targets behind walls and inside enclosed building structures [1]- [4]. However, there is still a need for producing high quality images that can effectively discriminate the targets of interest from clutter without increasing the data acquisition time.Several TWRI studies have focused on the use of polarization of the electromagnetic (EM) waves for detecting targets or enhancing the discrimination of targets [5]-[11]. In [5], a method for through-the-wall detection of certain types of weapons was developed by analyzing the...

Compressive-Sensing-Based High-Resolution Polarimetric Through-the-Wall Radar Imaging Exploiting Target Characteristics

“…Fig. 3(c) shows the sparse reconstruction result based on the method proposed in [5], which considers group sparsity across multiple polarization modes within the BCS framework. It is evident that the algorithm generates a low quality reconstruction result with many isolated and spurious pixels due to lack of consideration of the underlying target structure.…”

“…In group sparsity, the nonzero scattering coefficients have a common support for different polarizations, but their exact values may differ. The aformentioned properties have been separately considered in the context of TWRI to improve the sparse reconstruction performance, leading to higher imaging resolution, clutter reduction, and target separability [3]- [5].…”

Group Sparse Basis Pursuit Denoising Reconstruction Algorithm for Polarimetric Through-the-Wall Radar Imaging

“…Recently, several imaging algorithms based on compressive sensing (CS) technique have been developed for polarimetric TWRI system to reduce the polarimetric measurement data and enhance the imaging quality. In [6], the multitask Bayesian CS (MT-BCS) framework is employed to simultaneously reconstruct the images associated with all polarimetric channels and provide the enhanced image quality compared to the imaging results formed individually for each polarization channel. A modified clustered MT-BCS algorithm which combines the multipolarization sensing group sparsity and spatially clustered sparsity is proposed in [7] to achieve enhanced imaging capability for extended targets.…”

Wall clutter mitigation using HOSVD in through-the-wall radar imaging with compressed sensing

“…In [13], CS is firstly applied to recover the missing radar measurements and then a wall clutter mitigation method, using spatial filtering [17] or a subspace projection technique [18], is applied to remove the wall returns. In [14] and [15], sparse representation is used to first estimate the signal coefficients, and then singular value decomposition (SVD) is applied directly on the estimated coefficients in order to segregate the wall returns from the target signal. In the SVD-based wall clutter mitigation method [14], [15], [18], the wall and target subspaces are obtained from a two-dimensional data matrix, where one dimension represents the signal (or coefficient) index and the other represents the antenna index.…”

“…In [14] and [15], sparse representation is used to first estimate the signal coefficients, and then singular value decomposition (SVD) is applied directly on the estimated coefficients in order to segregate the wall returns from the target signal. In the SVD-based wall clutter mitigation method [14], [15], [18], the wall and target subspaces are obtained from a two-dimensional data matrix, where one dimension represents the signal (or coefficient) index and the other represents the antenna index. However, since the number of signal samples is often greater than the number of antennas, reducing the latter decreases the rank of the matrix, thus making it harder to separate the target subspace from the wall subspace.…”