Jiangjun Peng

High-dimensional 3D seismic data is inevitably contaminated by random noise due to environmental interference and acquisition system limitations. While many deep learning methods have been proposed for random noise attenuation, they often suffer from various limitations. For instance, some convolutional neural network (CNN)-based methods tend to be computationally inefficient when processing 3D seismic volumes, while implicit neural representation (INR)-based methods frequently fail to recover complex structural details owing to the spectral bias. To address these challenges, this study introduces an efficient nonlocal tensor function representation (NLTFR) for unsupervised 3D seismic denoising. Specifically, we cluster similar cubes within 3D seismic data into nonlocal similar groups. To efficiently and effectively exploit the nonlocal self-similarity inherent in seismic data, we employ tensor function factorization parameterized by INRs to model these clustered nonlocal similar groups. Furthermore, we propose a partially shared transfer learning strategy for NLTFR to accelerate model convergence when applied to large-scale seismic datasets. Multiple-frequency sinusoidal functions are incorporated into the INRs to mitigate spectral bias, thereby enhancing the model’s representation ability. Additionally, a weighted total variation and a second-order 3D total variation are introduced to improve denoising performance and robustness of the NLTFR model. Comprehensive experiments conducted on synthetic and field 3D seismic datasets demonstrate that the proposed NLTFR achieves a superior balance between efficiency and effectiveness compared with conventional deep learning methods. Overall, our method attains an average performance gain of 1.5dB while significantly reducing execution time relative to traditional deep learning methods for 3D seismic volumes, underscoring its practical efficiency.

Automated pavement distress classification using high-resolution Unmanned Aerial Vehicle (UAV) imagery is pivotal for intelligent transportation systems. However, long-term UAV monitoring faces a continuous stream of evolving distress types and changing remote sensing background textures, necessitating Class-Incremental Learning (CIL) capabilities. Existing methods struggle to balance stability and plasticity, especially under the severe storage limitations typical of local edge stations in air–ground collaborative systems. This data scarcity leads to catastrophic forgetting and confusion among fine-grained distress categories. To address these challenges, we propose a data-efficient approach named Analytic Geometric Alignment (AGA). Our framework mainly consists of three key components. First, to overcome the optimization gap between the feature extractor and the fixed geometric target, we introduce a Subspace-Aware Analytic Initialization (SAI) that computes a closed-form projection to instantly align the feature subspace with the ETF manifold before each task training. Second, on this aligned basis, a Decoupled Geometric Adapter (DGA) is incorporated to facilitate continuous non-linear adaptation to complex aerial textures. Finally, for stable incremental training, we design a Memory-Prioritized Regression (MPR) loss to enforce tighter geometric constraints on replay samples, significantly enhancing model stability. Extensive experiments on the UAV-PDD2023 dataset demonstrate that AGA significantly outperforms state-of-the-art methods, showcasing excellent robustness and data efficiency.

The recently proposed fully-connected tensor network (FCTN) decomposition has demonstrated significant advantages in correlation characterization and transpositional invariance, and has achieved notable achievements in multi-dimensional data processing and analysis. However, existing multi-dimensional data recovery methods leveraging FCTN decomposition still have room for further enhancement, particularly in computational efficiency and modeling capability. To address these issues, we first propose a FCTN-based generalized nonconvex regularization paradigm from the perspective of gradient mapping. Then, reliable and scalable multi-dimensional data recovery models are investigated, where the model formulation is shifted from unquantized observations to coarse-grained quantized observations. Based on the alternating direction method of multipliers (ADMM) framework, we derive efficient optimization algorithms with convergence guarantees to solve the formulated models. To alleviate the computational bottleneck encountered when processing large-scale multi-dimensional data, fast and efficient randomized compression algorithms are devised in virtue of sketching techniques in numerical linear algebra. These dimensionality-reduction techniques serve as the computational acceleration core of our proposed algorithm framework. Theoretical results on approximation error upper bounds and convergence analysis for the proposed method are derived. Extensive numerical experiments illustrate the effectiveness and superiority of the proposed algorithm over other state-of-the-art methods in terms of quantitative metrics, visual quality, and running time.

Hyperspectral anomaly detection (HAD) can be formulated as a binary classification problem, where the accurate separation of anomalies from the background relies on the precise features extraction or the effective prior representation of the background. Traditional methods based on hand-crafted regularizations, typically integrating low-rank and local smoothness priors of hyperspectral background images, offer strong interpretability. However, these methods rely on predefined regularization forms, which limit their ability to provide fine-grained, data-adaptive prior modeling for each specific instance. To overcome this limitation, we propose a novel model termed implicit Tucker subspace representation with sparsity regularization (ITSR- $\ell _{1}$ ). The motivation for ITSR- $\ell _{1}$ arises from the following three key points: 1) the low-rank Tucker decomposition can explicitly model the inherent low-rank structure of hyperspectral background images; 2) the core tensor obtained through Tucker decomposition tends to exhibit significant sparsity, which can be exploited further to enhance the modeling of the low-rank background component; and 3) the local smoothness of the original image can be transferred to the continuity of the column vectors of the factor matrices derived from Tucker decomposition. By parameterizing these vectors with implicit neural networks, ITSR- $\ell _{1}$ enables data-adaptive, fine-grained characterization of continuity priors across the background.

For infrared small target detection tasks, deep unfolding techniques have demonstrated effectiveness and practical value. However, existing methods generally emphasize the low-rankness of background and the sparsity of targets within the robust principal component analysis (RPCA) framework, which may overlook the intrinsic gradient prior information existed in background. To address the challenges of complex background estimation and accurate small target detection, we propose a gradient prior-guided deep unfolding network, termed the correlated total variation network (CTVNet). First, we introduce a correlated total variation regularization to simultaneously characterize the low-rankness and local smoothness of the background, and transform it into the estimation of gradient maps. Subsequently, we employ a multi-scale feature fusion network to thoroughly extract gradient priors, replacing the complex and limited analytical computation of gradient correlations. Finally, we unfold the designed iterative algorithm using alternating direction method of multipliers (ADMM) into a learnable network, where each module corresponds to a specific operator within the iterative process, and all parameters are learnable. By training the network end-to-end, the learnable modules can be automatically optimized to better separate the background and the target. Extensive experimental results demonstrate that our proposed method achieves competitive performance compared to several state-of-the-art algorithms while exhibiting superior performance and generalization capabilities on both in-distribution and out-of-distribution data. Our code is available at https://github.com/AuroraPei/CTVNet.