Non-negative matrix factorization based approaches for wall mitigation in TWRI

Kumlu, Deniz; Erer, İşın; Paker, Selçuk

doi:10.1007/s11760-021-02031-z

Cited by 5 publications

(4 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The NMF technique uses a different decomposition methodology than previous subspace methods to generate a low‐rank approximation of the data matrix. In NMF, the product of two matrices W and H can approximate the data matrix R , where both matrices must be nonnegative and so can be written as 12

R\approx WH\,\text{suchthat}\,W,H\ge 0.

…”

Section: Clutter Reduction Methodsmentioning

confidence: 99%

“…By setting the rank K < N , a low‐rank approximation of the data matrix X can be obtained. Because the clutter's response is far greater than the target's, the clutter can be regarded as the most significant decomposition component, and it can be obtained by setting K = 1 12

{R}_{clutter}={W}_{M\times 1}{H}_{1\times N},

and R target can be easily obtained by

{R}_{t\text{arg}et}=X-{X}_{clutter}.

…”

Section: Clutter Reduction Methodsmentioning

confidence: 99%

“…Subspace methods have two notable limitations: the need for accurate manual selection of ranks corresponding to clutter and target subspaces and susceptibility to failure in the presence of substantial random noise. The decomposition into low‐rank (L) and sparse matrix (S) frameworks presents a potential solution to these challenges 11,12 . The regularization parameter plays a pivotal role in controlling the trade‐off between the sparse component (target) and the low‐rank component (clutter and noise).…”

Section: Introductionmentioning

confidence: 99%

“…The decomposition into low-rank (L) and sparse matrix (S) frameworks presents a potential solution to these challenges. 11,12 The regularization parameter plays a pivotal role in controlling the trade-off between the sparse component (target) and the low-rank component (clutter and noise). Adjusting the regularization parameter value allows for fine-tuning the decomposition outcome.…”

mentioning

confidence: 99%

See 3 more Smart Citations

Performance analysis of signal processing algorithms for wall clutter mitigation and contrast target detection

Singh

2024

Micro & Optical Tech Letters

View full text Add to dashboard Cite

The presence of clutter in through‐the‐wall images significantly degrades image quality and hampers the performance of algorithms used for data processing in tasks such as target detection, identification, or reconstruction. Real building walls exhibit inherent inhomogeneities, featuring varying frequency and spatial properties that defy the assumption of a smooth surface. Consequently, they produce nonuniform wall surface reflections at each scanning position. Furthermore, these walls often incorporate supply pipes that introduce substantial clutter, which, although stronger than the target responses, remains weaker than the wall surface clutter. Additionally, this clutter often exhibits signatures similar to those of the target. In real‐world scenarios, target reflections can manifest as wide, flat hyperbolas or nearly straight lines, and they may be positioned near the wall. Consequently, distinguishing targets from clutter becomes a complex challenge. Various clutter reduction methods have been proposed in recent years, showing varying degrees of success. However, the effectiveness of these methods in the literature is inconsistent and sometimes contradictory. To address this issue, a comprehensive investigation of well‐established clutter reduction methods was conducted to evaluate their performance under identical conditions. These methods were rigorously assessed using practical radar‐measured data acquired in the presence of actual building wall materials and contrasting targets. Evaluation criteria include the target‐to‐clutter ratio and peak signal‐to‐noise ratio. The results of the experiment revealed that independent component analysis outperformed other clutter reduction methods, demonstrating superior performance in mitigating clutter and enhancing target detection.

show abstract