Search Results (98)

Compressed sensing (CS) and matrix completion algorithms (MCA) have each introduced sparse and low-rank priors into synthetic aperture radar (SAR) imaging. However, their combined use reveals a fundamental zero-sum trade-off: enhancing spatial continuity tends to obscure weak targets, while strengthening sparse recovery amplifies off-grid artifacts. This inherent conflict is further exacerbated by static regularization, which imposes a rigid global compromise and prevents genuine synergy between the two priors. To overcome this limitation, this paper proposes a Spatio-Temporal Adaptive Synergistic Constraint Imaging (STASCI) algorithm, which dynamically balances the two priors in a scene-aware manner. The core of STASCI is a unified regularization framework. The low-rank constraint models’ spatial continuity in the background to suppress off-grid artifacts. The sparse constraint, enhanced by a non-convex Geman-McClure function, is employed to detect weak targets and compensate for detail loss. A key innovation is a spatio-temporal dual-dimensional regularization mechanism that employs Sobel operators to probe local spatial gradients and dynamically adjusts the strength of each prior according to regional scene characteristics. This enables adaptive synergy rather than a fixed trade-off. The optimization is solved via the alternating direction method of multipliers (ADMM), with the low-rank subproblem accelerated by randomized singular value decomposition (RSVD). Final imaging is performed using the Range Migration Algorithm (RMA). Experiments on real measurements and public datasets demonstrate that STASCI breaks the conventional detail-background trade-off. It effectively suppresses off-grid artifacts while retaining weak targets, leading to significant improvements in imaging accuracy and robustness across complex scenarios. Full article

Accurate characterization of coherent vortex structures in high-speed turbulent boundary layers presents a persistent challenge due to the flow’s high dimensionality and nonlinear dynamics. This study investigates an optimized decomposition framework that integrates modal decomposition techniques with a novel vortex identification strategy to extract dynamically significant features. The numerical solution from a previously conducted high-fidelity simulation of MVG-controlled supersonic flow serves as the testbed. Principal Component Decomposition and Non-negative Matrix Factorization are applied across multiple flow variables to evaluate their effectiveness in isolating coherent structures. The results show that, across the velocity-based cases, 3–4 modes capture 70% of the TKE with MSE about 0.1, while the Liutex case requires 14 modes but achieves a lower MSE of about 0.04. Overall, using the same number of modes yields similar reconstruction performance across all cases. The influence of various normalization and rescaling methods on decomposition performance is also examined. Optimization is guided by two primary criteria: the interpretability of spatial modes and MSE in reconstructing vortex structures. By employing low-rank matrix representations, this optimization study aims to enhance interpretability and reduce computational costs. This approach establishes a mathematically rigorous and efficient platform for analyzing vortex dynamics, achieving significant dimensionality reduction while preserving key features of turbulent transport. Full article

Vision–language models (VLMs) are increasingly deployed in resource-constrained environments, yet efficient fine-tuning remains challenging because post-training quantization often degrades the effectiveness of low-rank adaptation. This paper revisits that mismatch in the context of MobileVLM1.7B and presents QuantFT-VL, a novel initialization strategy following the quantization phase to seamlessly align with the LoRA technique. The key idea is to initialize LoRA using a low-rank approximation of the quantization residual instead of the default zero-initialization used in QLoRA-style pipelines. After quantizing a pretrained weight matrix W into Q, we compute the residual W − Q and use truncated singular value decomposition to initialize the LoRA factors (A and B) so that the starting adapted weight Q + AB^T better matches the full-precision model. This residual-aware initialization reduces the discrepancy introduced by quantization and leads to faster and more stable optimization. Experiments on six standard VLM benchmarks show that QuantFT-VL consistently improves over QLoRA and recovers performance close to or better than full-precision LoRA in the best setting. On two RTX 3090 GPUs, QuantFT-VL improves the average benchmark score by 3.27 percentage points over QLoRA while preserving the memory and speed advantages of quantized fine-tuning. Full article

Deep learning (DL) methods have achieved promising performance in infrared weak target segmentation. However, their interpretability and robustness against cluttered backgrounds and noise remain limited. We propose an adaptive-weighted deep unfolding network (AwDuNet) that unfolds alternating direction method of multipliers (ADMM) iterations for adaptive sparse–low-rank decomposition into multi-stage interpretable modules for end-to-end training. An adaptive weight matrix is jointly estimated from a local structural-difference matrix and a sparse-enhancement matrix, thereby strengthening target–background separation while preserving fine target details. To suppress background clutter, we design a dual-path complementary attention (DCA) mechanism for the low-rank background reconstruction module (LBRM), which improves low-rank background modeling by jointly leveraging spatial and channel attention. By extracting local details and global context in parallel, DCA enhances weak-target responses and mitigates interference from complex backgrounds. We also build a real-scene infrared dataset with 632 images for out-of-domain evaluation. The model is tested without fine-tuning after training on public datasets to assess practical robustness. Experiments on multiple public datasets validate the effectiveness and generalization of AwDuNet. Full article

Low-rank sparse decomposition models have become the mainstream optimization framework for multiframe infrared small target detection. Existing low-rank matrix decomposition approximations typically pre-decompose infrared videos into the product of two low-rank matrices to capture the background’s low-rank characteristics. However, such approximations are not optimal and often result in suboptimal background recovery. To achieve more accurate low-rank recovery, we exploit the intrinsic relationship between low-rank matrices and their generalized inverse matrices, thereby improving conventional decomposition approximations. Moreover, to address the high computational cost of applying low-rank and sparse decomposition models to multi-frame infrared videos, we introduce a robust column-row (CUR) decomposition to accelerate the iterative process, thereby significantly improving computational efficiency. The experimental results show that the proposed method achieves fast detection of small targets in infrared videos while maintaining competitive detection performance. Full article

Polarimetric information is essential for scattering interpretation and target characterization in synthetic aperture radar (SAR) remote sensing, yet many resource-constrained platforms (e.g., small satellites and unmanned aerial vehicles (UAVs)) operate with limited polarization modes or even a single radio frequency (RF) chain, which limits full polarimetric scattering acquisition. To address this limitation, this paper proposes a single-channel framework for estimating the full polarization scattering matrix (PSM) enabled by time-varying polarization modulation. The transmit/receive polarization states are steered along predefined trajectories on the Poincaré sphere to generate time-varying polarization tags that are encoded into the received echoes through the target’s polarization-varying response. A compact observation model is then derived to relate the single-channel echoes, the known polarization tags, and the unknown PSM; based on this, the PSM is then estimated via a least squares formulation with a low-rank approximation. Simulation results demonstrate the robust reconstruction of the full polarimetric scattering matrix under diverse modulation trajectories. For arbitrarily chosen random point targets, when the signal-to-noise ratio (SNR) exceeds −20 dB, the polarimetric similarity coefficient approaches 1, and the estimation errors of Pauli power components converge toward zero. Furthermore, the method’s reliability is validated on distributed vegetation clutter. Quantitative metrics demonstrate near-perfect statistical consistency, with polarimetric entropy and alpha angle errors within 0.14%. Overall, the proposed approach provides a practical pathway to enhance the availability of full polarimetric scattering information under limited-observation conditions, confirming its feasibility for downstream analysis in complex natural scenes while maintaining a single radio frequency (RF) chain architecture augmented by a polarization modulator. Full article

Tensor completion is an extension of matrix completion aimed at recovering a multiway data tensor by leveraging a given subset of its entries (observations) and the pattern of observation. The low-rank assumption is key in establishing a relationship between the observed and unobserved entries of the tensor. The low-rank tensor completion problem is typically solved using numerical optimization techniques, where the rank information is used either implicitly (in the rank minimization approach) or explicitly (in the error minimization approach). Current theories concerning these techniques often study probabilistic recovery guarantees under conditions such as random uniform observations and incoherence requirements. However, if an observation pattern exhibits some low-rank structure that can be exploited, more efficient algorithms with deterministic recovery guarantees can be designed by leveraging this structure. This work shows how to use only standard linear algebra operations to compute the tensor train decomposition of a specific type of “fiber-wise” observed tensor, where some of the fibers of a tensor (along a single specific mode) are either fully observed or entirely missing, unlike the usual entry-wise observations. From an application viewpoint, this setting is relevant when it is easier to sample or collect a multiway data tensor along a specific mode (e.g., temporal). The proposed completion method is fast and is guaranteed to work under reasonable deterministic conditions on the observation pattern. Through numerical experiments, we showcase interesting applications and use cases that illustrate the effectiveness of the proposed approach. Full article

The deployment of transformer-based language models on resource-constrained edge devices presents fundamental challenges in computational efficiency and memory utilization. We introduce SQ-LoRA (Stable-rank Quantized Low-Rank Adaptation), a theoretically grounded compression framework that achieves unprecedented efficiency through the synergistic integration of adaptive low-rank decomposition, hardware-accelerated structured sparsity, and intelligent hybrid quantization. Our primary contribution establishes the first rigorous mathematical connection between the matrix stable rank and optimal LoRA rank selection, formalized in Theorem I, which provides bounded approximation guarantees. SQ-LoRA implements: (1) adaptive rank allocation via stable-rank analysis to automatically determine layer-wise compression ratios; (2) 4:8 structured sparsity patterns, enabling 2× hardware acceleration on modern edge processors; and (3) a three-tier quantization scheme that combines 4-bit NormalFloat storage with selective 3-bit/8-bit precision to preserve outliers. A comprehensive evaluation on four diverse natural language processing (NLP) benchmarks demonstrates that SQ-LoRA achieves a 320 MB memory footprint (96.7% reduction) and a 10 ms inference latency (91.7% improvement), and maintains 82.0% average accuracy (within 0.15% of the full model). Statistical significance testing (p < 0.001) confirms its superiority over state-of-the-art methods. This framework enables the deployment of sophisticated language models on devices with 2 GB of RAM, advancing practical edge-AI applications. Full article

Infrared small target detection (IRSTD) task is vital in practical applications. It is still a challenge when the target size is very small and the local signal-to-noise ratio is particularly low. This paper proposed an Infrared Tall Patch-Matrix (ITPM) model, which takes a novel perspective to construct a lower-rank patch matrix structure to improve the detection performance of low-contrast small targets. Specifically, we use a sliding split window to reconstruct the original image into a suitable low-rank structure called Tall Patch-Matrix, which can increase the detection rate of low-contrast small targets and suppress most noise. Second, the High Local Variance Low-Rank and Sparse Decomposition (ITPM-HiLV-LRSD) method is used to perform low-rank and sparse decomposition of the Infrared Tall Patch-Matrix, and a Thin Singular Value Decomposition (Thin SVD) optimization strategy is proposed to further reduce the computational complexity. Given the absence of open literature datasets for detecting infrared targets in low-contrast small scenarios, we created a Low-contrast Small Target Detection Dataset (LSTDD) comprising 600 infrared target detection images with varied sky backgrounds. This dataset simulates low-contrast small targets across different signal-to-noise ratios. To demonstrate the generalizability of our method, we also conducted experiments on a representative low-contrast subset of real-world images from the SIRST dataset. Compared with six state-of-the-art methods, our proposed method excels in detecting low-contrast small targets with superior performance. Full article

With the continuous development of power systems, WAMS have become increasingly important for real-time system monitoring. As the core devices of WAMS, PMUs can provide synchronized, high-precision, and high-resolution measurements of power system states. However, in practical applications, PMUs deployed in different regions often operate at different sampling rates, resulting in multi-rate measurement data and posing challenges for data fusion. To address this issue, this paper proposes a multi-rate PMU data fusion method based on low-rank TT. Specifically, the proposed method first performs tensor-based modeling of multi-rate measurement data, embedding multidimensional correlations into a high-order tensor representation. Then, a data completion model is constructed through low-rank TT decomposition to effectively capture cross-timescale dependencies. Finally, an efficient numerical solution is developed to expand low-resolution measurements into high-resolution data, thereby achieving unified data fusion. Case studies on both simulated and real-world PMU measurement data demonstrate that the proposed approach outperforms traditional interpolation and matrix completion methods, achieving superior reconstruction accuracy and robustness. Full article

Satellite videos are widely applied for large-scale surveillance. Existing low-rank matrix decomposition-based methods produce promising results under simple and stationary backgrounds. However, these methods suffer a severe performance drop on satellite videos with complex and dynamic backgrounds. To address these challenges, we propose a matrix-based total variation regularized robust PCA (TV-RPCA) approach for moving-vehicle detection. Specifically, our TV-RPCA uses the partial sum of singular values to model the low-rank background. Moreover, a ${\mathit{\ell }}_p$ norm and a spatial–temporal TV regularization are adopted to encourage the spatial–temporal continuity of foregrounds. The optimization of our TV-RPCA is carried out using the augmented Lagrangian multiplier framework combined with the alternating direction minimization approach. Comprehensive experiments conducted on SkySat and Jilin-1 video data verify the effectiveness of the proposed approach. Full article

A long-standing challenge for the structural health monitoring (SHM) community is the masking effect of environmental variability, typically addressed by orthogonal projection (OP)-based data normalization to isolate the influence of environmental variability and enable structural anomaly detection. However, conventional OP techniques, such as principal component analysis, rely on clean and complete data, and their performance degrades in the presence of outliers or missing entries. To overcome this limitation, this paper proposes an integrated approach that combines OP with noisy low-rank matrix completion (NLRMC). The main advantage of NLRMC is its ability to couple low-rank and sparse decomposition with matrix completion, simultaneously handling data corruption and missingness to recover incomplete datasets and enable robust anomaly detection. By incorporating novelty-indicator extraction, a fully online, unsupervised anomaly-detection procedure is established. Validation on a vibration-based SHM dataset from the KW51 railway bridge confirms that the NLRMC-OP approach achieves reliable detection of operational state changes before and after retrofitting, even under both data corruption and missing scenarios. This study advances the usability of SHM data and facilitates efficient decision-making, while also highlighting the broader significance of leveraging the low-rank data structure in AI-enabled operation and maintenance of civil infra-structure. Full article

Mainlobe interference can severely degrade the performance of distributed phased-array radar systems in the presence of strong jamming or low-reflectivity targets. This paper introduces a signal–data dual-domain cooperative antijamming and localization (SDCAL) framework that integrates adaptive complete ensemble empirical mode decomposition with improved blind source separation and wavelet optimization (CEEMDAN-WOBSS) for signal-level denoising and separation. Following source separation, CFAR-based pulse compression is applied for precise range estimation, and multi-node data fusion is then used to achieve three-dimensional target localization. Under low signal-to-noise ratio (SNR) conditions, the adaptive CEEMDAN–WOBSS approach reconstructs the signal covariance matrix to preserve subspace rank, thereby accelerating convergence of the separation matrix. The subsequent pulse compression and CFAR detection steps provide reliable inter-node distance measurements for accurate fusion. The simulation results demonstrate that, compared to conventional blind-source-separation methods, the proposed framework markedly enhances interference suppression, detection probability, and localization accuracy—validating its effectiveness for robust collaborative sensing in challenging jamming scenarios. Full article

Hyperspectral image (HSI) denoising is an important preprocessing step for downstream applications. Fully characterizing the spatial-spectral priors of HSI is crucial for denoising tasks. In recent years, denoising methods based on low-rank subspaces have garnered widespread attention. In the low-rank matrix factorization framework, the restoration of HSI can be formulated as a task of recovering two subspace factors. However, hyperspectral images are inherently three-dimensional tensors, and transforming the tensor into a matrix for operations inevitably disrupts the spatial structure of the data. To address this issue and better capture the spatial-spectral priors of HSI, this paper proposes a modeling approach named low-rank Tucker decomposition with subspace implicit neural representation (LRTSINR). This data-driven and model-driven joint modeling mechanism has the following two advantages: (1) Tucker decomposition allows for the characterization of the low-rank properties across multiple dimensions of the HSI, leading to a more accurate representation of spectral priors; (2) Implicit neural representation enables the adaptive and precise characterization of the subspace factor continuity under Tucker decomposition. Extensive experiments demonstrate that our method outperforms a series of competing methods. Full article

Transformer architecture, initially developed for natural language processing and time series analysis, has been successfully adapted to various generative models in several domains. Object pose estimation, which uses images to determine the 3D position and orientation of an object, is essential for tasks such as robotic manipulation. This study introduces a transformer-based deep learning model for object pose estimation in computer vision, which determines the 3D position and orientation of objects from images. A baseline model derived from an encoder-only transformer faces challenges with high GPU memory usage when handling multiple objects. To improve training efficiency and support multi-object inference, it reduces memory consumption by adjusting the transformer’s attention layer and incorporates low-rank weight decomposition to decrease parameters. Additionally, GQA and RMS normalization enhance multi-object pose estimation performance, resulting in reduced memory usage and improved training accuracy. The improved model implementation with an extended matrix dimension reduced the GPU memory usage to only 2.5% of the baseline model, although it increased the number of model weight parameters. To mitigate this, the number of weight parameters was reduced by 28% using low-rank weight decomposition in the linear layer of attention. In addition, a 17% improvement in rotation training accuracy over the baseline model was achieved by applying GQA and RMS normalization. Full article