Schrödinger Bridge-based diffusion models have demonstrated promising performance in signal denoising. However, since ground truth signals are unavailable during the sampling process, neural networks must be employed to learn the mapping, which breaks the theoretical coupling between diffusion and sampling processes. This paper reveals a critical inconsistency between the theoretical diffusion path and the learned sampling trajectory across different frequency bands. This diffusion-sampling inconsistency directly undermines denoising effectiveness. To address this limitation, we propose the Frequency-Dependent Scheduled Schrödinger Bridge (FDSSB), which leverages power spectral density to adaptively schedule diffusion processes across frequencies. This mechanism assigns asynchronous diffusion schedules to different frequency components, correcting the diffusion schedule to better match the sampling process. As a result, FDSSB effectively mitigates the mismatch and enhances the consistency between diffusion and sampling processes. Extensive experiments demonstrate that FDSSB achieves state-of-the-art performance, with an average scale-invariant signal-to-noise ratio improvement of 7.9066 dB over competitive approaches.

Although salient object detection (SOD) methods inspired by human attention mechanisms have received increasing attention for their superior performance, they suffer from a trade-off between efficacy and efficiency, due to the high performance generally coming at the expense of large parameter sizes and high computational costs. To address this issue, we propose a Collaborative Prior-Enhanced RGB-D Salient Object Detection Network (CPENet) for Intelligent IoT Perception Devices. Specifically, we propose a dual-stream architecture utilizing MobileViT as the backbone to tackle the challenge of capturing long-range dependencies among multi-modal features with reduced model complexity. Then, a Scale Information Enhancement Module (SIEM) is proposed to effectively enhance and fuse multi-scale features. To adaptively aggregate multi-modal features, we propose a Prior-Driven Modality Aggregation (PDMA) that integrates the coarse saliency maps (Prior-Glance) of model into cross-modal fusion. Furthermore, we propose an extremely simple shared decoder to both predict the saliency maps of fused features for deep supervision and generate Prior-Glance in generation modules. Finally, considering the deficiency of boundary representation, unlike other methods which append costly edge enhancement module, a prior-driven region loss is designed that improves the capacities of boundary representation and generalization of the model under the guidance of Prior-Region. Extensive experiments demonstrate that the proposed CPENet outperforms most state-of-the-art methods on seven datasets with fewer parameters (15.7M) and lower computational complexity (21.9GFLOPs). Codes and results are available at https://github.com/blossom-lv/CPENet.

ABSTRACT Visual object tracking is widely applied in intelligent transportation systems and visual surveillance systems that serve smart cities, as well as in autonomous vehicles. Existing methods usually utilise a relation‐modelling framework to model the visual object tracking problem, with auxiliary spatial context and temporal information. The spatial context is often extracted by enlarging the target template, which can introduce more background and positional information. The temporal correlation is obtained by associating the search image with previous images. However, due to noise interference, existing methods often partially exploit auxiliary data, leading to underutilisation of spatiotemporal information. To address these issues, we propose a novel and concise tracking framework, uniformly encoding all auxiliary data, including the enlarged target template, previous images, and corresponding target bounding boxes. Specifically, to mitigate the unstable factors introduced by these raw inputs, we propose a spatiotemporal context adaptive encoder, which can adaptively select appropriate information in noisy data. Extensive experiments show that the proposed method achieves state‐of‐the‐art performance on various benchmarks, demonstrating its superiority.

In the field of autonomous driving, the micro-electromechanical systems (MEMS)-based vehicle navigation usually adopts multi-sensor integrated navigation to achieve high-precision positioning. However, due to the complex environments, the accuracy and reliability of navigation sensors may be significantly reduced. To address these challenges, the interacting multiple model (IMM)-based strapdown inertial navigation system/global navigation satellite system/odometer/non-holonomic constrain (SINS/GNSS/OD/NHC) integrated navigation is adopted. Unfortunately, due to the use of traditional state-space model (SSM), the existing IMM estimation methods often suffer from poor estimation consistency, and the mounting error angle will also make OD/NHC subfilter models affect estimation consistency during the interaction process. Moreover, complex environments lead to frequent model switching, and relying on inaccurate model probabilities may cause significant fluctuations in the subfilter outputs, thereby reducing estimation accuracy. In contrast, the proposed IMM estimation method constructs Lie group-based subfilter SSM, which improves estimation consistency. Additionally, the velocity-bias-based mounting error angle estimation method is proposed by using variational Bayesian (VB) techniques, which further refines the OD/NHC models in IMM. On the other hand, a dynamic likelihood adaptive mechanism (DLAM) is introduced to improve reliability and mitigate the negative effects of frequent switching. Simulation and field test results demonstrate that the proposed velocity-bias-based mounting error angle estimation method has fast convergence speed and high convergence accuracy. Additionally, the proposed Lie group-based reliable IMM estimation method has better accuracy and robustness in complex environments as compared to the existing IMM estimation methods.

The absorption and scattering of light caused by water and suspended particles in the ocean may lead to underwater image degradation. Existing deep learning-based underwater image enhancement models suffer from an excessive number of parameters, which makes them unsuitable for deployment on resource-constrained underwater Internet of Things (IoT) devices. Furthermore, the evaluation methods for enhanced results may not consider the feature richness. In this work, a novel lightweight underwater image enhancement framework and three baseline models based on the framework are proposed. The proposed framework is inspired by the atmospheric imaging model and histogram differences, and it believes that the reference image can be expressed as a linear combination of the degraded image and two nonlinear transformations, termed as the compensation map and the suppression map. This framework compels the neural network model to use the learned parameters to represent the nonlinear relationships among the degraded image, compensation map, and suppression map, rather than to generate high-level semantic information within an encoder-decoder framework. As a result, the number of parameters is significantly reduced. This paper also proposes a novel subjective evaluation method for underwater images, termed as Feature Entropy Evaluation (FEE). The proposed FEE method outperforms existing evaluation methods for underwater scenes with color checker plates by delivering rankings that more accurately capture the absolute value differences in the color checker plates. Extensive experiments on public datasets demonstrate that, as compared to other existing methods, the proposed models achieve comparable results with only 0.01% parameters of existing methods. The relevant code will be made available after the paper is accepted.

Hyperspectral image change detection (HSIs-CD) often faces the challenge of limited sample sizes, and labeling data is both time-consuming and labor-intensive. Foundation models leverage extensive unlabeled data for self-supervised generative pre-training, allowing the model to learn rich data representations. However, few models have been specifically designed for HSIs, and existing methods often rely on pre-training datasets that are limited to data from a small number of satellite sensors. This limitation affects generalization, especially when there are significant differences between data from different sensors. Moreover, the difference map (DMP) of bi-temporal HSIs is often used as input to the networks. While the DMP-based approach reduces FLOPs, it may lead to information loss compared to dual-branch networks. In this letter, we propose a mini-patch-based generative pre-trained spectral-spatial transformer (GPSST) for semi-supervised HSIs-CD. We begin by collecting public HSIs datasets and dividing them into thousands of patches. Each patch is then split into spectral-spatial tokens, with a portion of these tokens masked and used as input for the GPSST. We then design a spectral-spatial masked autoencoder (MAE) as the backbone of GPSST for self-supervised generative learning. Finally, we fine-tune the GPSST encoder using a small number of labeled patches and design a principal component analysis (PCA) branch to compensate for the information loss caused by the DMP. Our experiments demonstrate that GPSST outperforms existing methods, achieving superior accuracy in HSIs-CD.

The unknown process noise covariance matrix (PNCM) problem inducing by poor calibration or time-varying environment has not been addressed in the 2-D multi-robot system. This problem will severely deteriorate the distributed cooperative localization consistency and accuracy, and is troublesome to solve due to small magnitude of the 2-D robot’s PNCM. In this paper, the above issue is addressed by the following two steps. Firstly, the motion model of the 2-D robot is reconstructed to form a more estimable PNCM, from which a small-scale PNCM estimation algorithm is derived. Then the cooperative strategy consisting of a Kullback-Leibler average strategy and a recovery strategy is proposed to guarantee global PNCM estimation consensus and convergence, even if only partial robots access absolute measurement information. Theoretical consensus and convergence analyses are presented and comprehensive simulation and experimental tests are conducted to verify the effectiveness and superiority of the proposed algorithm. Note to Practitioners—This work is motivated by the inaccurate PNCM problem existing in the 2-D homogeneous multi-robot system, whose PNCM is very small in magnitude. The pose accuracy of the 2-D mobile robot, which relies on the sensor precision, is generally not enough to estimate such small-scale PNCM. Most of the existing PNCM estimation algorithms are regarding to simple target tracking models whose PNCMs are relatively large in magnitude. Furthermore, a few small-scale PNCM estimation algorithms make crucial assumptions about the PNCM, which limits their practicality. This paper proposed a novel small-scale PNCM estimation algorithm and an efficient cooperative strategy to facilitate global PNCM estimation consensus and convergence, without making any assumptions about the PNCM. The consensus and convergence analyses are provided to further demonstrate the effectiveness of the proposed adaptive cooperative localization algorithm. The proposed algorithm has been evaluated via simulation, public dataset and physical experiment.

This article proposes a novel recognition-guided denoising method for underwater acoustic signals, termed RGDNet, which is designed to bridge the gap where existing methods, despite achieving high denoising metrics, fail to effectively enhance recognition accuracy. Specifically, we first designed a task-specific mask integration (TSMI) module that minimizes phase and frequency distortions by converting signal fusion across tasks into a unified mask-based fusion, ensuring the fusion of signal features remains effective. We further developed a context-aware signal decoupling (CASD) module to maintain the distinctiveness of sub-task signals through segmented context analysis. Following this, a recognition-guided branch (RGB) equipped with an auxiliary classifier is proposed, which guides the optimization of the denoising process based on recognition loss. Additionally, we also implemented a complex domain multimetric discriminator (CDMD) that assesses the quality of denoised signals from both metric-focused and task-specific perspectives, effectively linking denoising metrics with recognition performance. Extensive experiments show that RGDNet significantly outperforms existing methods, increasing the scale-invariant signal-to-noise ratio (SI-SNR) by an average of 6.33 dB and enhancing recognition accuracy by 11.26%.

Decentralized cooperative localization (DCL) exhibits significant advantages in fault tolerance, practicality, and scalability, which serves as a crucial prerequisite for multi-robot system to achieve effective cooperative operations. Unfortunately, sensor measurements inevitably contain outliers due to the high dynamics of robots and uncertain environmental factors in practical applications. Most of the existing DCL algorithms concentrated on studying the cross-correlation between estimates, and performed poorly in the extreme cases with measurement outliers. To enhance the robustness and stability of multi-robot system, a robust DCL (RDCL) framework is proposed to suppress the impacts of unknown sensor outliers. We improve the measurement update process of the traditional DCL algorithm by employing two outlier-robust extend Kalman filter (EKF) methods to adaptively fuse outlier-contaminated measurements based on tracking the correlation between robots, which thereby achieves accurate and robust localization. The proposed RDCL framework is not restricted to a specific model and inhibitory effects on both absolute and relative measurements outliers. Simulation and experimental results demonstrate the potential and advantages of the proposed algorithm in terms of accuracy, robustness, and stability for multi-robot cooperative localization.

Supervised deep learning (DL)-based hyperspectral images change detection (HSIs-CD) has demonstrated excellent performance; however, current methods require many labeled training samples, and labeling the dataset is labor-intensive, limiting the application of high-precision supervised learning. Besides, there has been a lack of breakthroughs in unsupervised HSIs change detection (CD) methods due to the different feature distributions of bitemporal HSIs. Here, we propose a semi-supervised domain alignment transformer (DA-Former) for HSIs-CD to address the issues with limited samples. Specifically, a dual-branch transformer autoencoder (TAE) is designed, where the middle layer weights of the dual-branch transformer are shared, pulling features from different data into the same space. Moreover, the TAE is also trained cyclically to align the domains. Although the bitemporal HSIs features are cross domain, there is still confusion between the features of different objects. Thus, two fully connected (FC) layers are employed to classify the HSIs middle features extracted by the TAE into changed class or unchanged class with limited labeled data. Three HSIs-CD datasets are used to test this method, showing that TAE can align bitemporal HSIs domains and achieve the highest accuracy compared with benchmark approaches. The code of the proposed method will be published athttps://github.com/yanhengwang-heu/IEEE_TGRS_DA-Former.

This paper presents an underwater acoustic target recognition method to reduce recognition errors in continuous recordings caused by variations in ship operating conditions. The proposed method comprises two-stages: the spectral feature classification and the supervised contrastive learning, and it is called as SFC-Sup as a result in this paper. In the first stage, a new spectral feature classification strategy is designed to choose appropriate feature sets for contrastive learning, based on which an instance discrimination pretext task is created by utilizing different spectral features to capture invariant features across segments under different operating conditions. In the second stage, a dynamic weighted loss function is introduced to guide the joint optimization process in the framework of contrastive learning. Different to existing methods which focus on improving the recognition accuracy by designing features for individual segments, the proposed two-stage method SFC-Sup considers consistent features across diverse segments, which is expected to improve recognition accuracy in a continuous recording. Experimental results demonstrate that in the presence of complex operating conditions, SFC-Sup exhibits superior stability and enhances recognition accuracy by 2.06% compared to state-of-the-art methods.

This paper presents a novel multiple-outlier-robust Kalman filter (MORKF) for linear stochastic discretetime systems. A new multiple statistical similarity measure is first proposed to evaluate the similarity between two random vectors from dimension to dimension. Then, the proposed MORKF is derived via maximizing a multiple statistical similarity measure based cost function. The MORKF guarantees the convergence of iterations in mild conditions, and the boundedness of the approximation errors is analyzed theoretically. The selection strategy for the similarity function and comparisons with existing robust methods are presented. Simulation results show the advantages of the proposed filter.

Convolutional neural networks (CNNs) with excellent spatial feature extraction abilities have become popular in remote sensing (RS) image change detection (CD). However, CNNs often focus on the extraction of spatial information but ignore important spectral and temporal sequences for hyperspectral images (HSIs). In this paper, we propose a joint spectral, spatial, and temporal transformer for hyperspectral image change detection (HSI-CD), named SST-Former. First, the SST-Former position-encodes each pixel on the cube to remember the spectral and spatial sequences. Second, a spectral transformer encoder structure is used to extract spectral sequence information. Then, a class token for storing the class information of a single temporal HSI concatenates the output of the spectral transformer encoder. The spatial transformer encoder is used to extract spatial texture information in the next step. Finally, the features of different temporal HSIs are sent as the input of temporal transformer, which is used to extract useful CD features between the current HSI pairs and obtain the binary CD result through multilayer perception (MLP). We evaluate SST-Former on three HSI-CD datasets by numerous experiments, showing that it performs better than other excellent methods both visually and qualitatively.

In this article, a novel normal-skew mixture (NSM) distribution is presented to model the normal and/or heavy-tailed and/or skew nonstationary distributed noises. The NSM distribution can be formulated as a hierarchically Gaussian presentation by leveraging a Bernoulli distributed random variable. Based on this, a novel robust Kalman filtering framework can be developed utilizing the variational Bayesian method, where the one-step prediction and measurement-likelihood densities are modeled as NSM distributions. For implementation, several exemplary robust Kalman filters (KFs) are derived based on some specific cases of NSM distribution. The relationships between some existing robust KFs and the presented framework are also revealed. The superiority of the proposed robust Kalman filtering framework is validated by a target tracking simulation example.

This paper investigates the state estimation error covariances of cyber-physical systems (CPSs) when its physical system and transmission channels are attacked by false data injection (FDI). Specifically, sensors in CPSs measure the physical systems' state and then send these measurement information to a fusion center through transmission channels. Due to easy deployment and maintenance of sensors as well as unprotected wireless transmission medium, CPSs in sensing and communication are vulnerable to malicious attacks. To maliciously degrade the system estimation performance, an intruder attempts to modify the state and measurements by launching FDI attacks. To seek an optimal attack strategy from the perspective of an attacker, the corresponding estimate error covariances of CPSs are calculated. Based on this, we compare the estimation error covariances of different attacks on the physical system and transmission channels. Simulations demonstrate the effectiveness of the theoretical results.

Abstract This paper proposes a novel secure diffusion Kalman filter (dKF) algorithm to improve the estimation performance crippled by false data injection attacks on sensors in wireless sensor networks (WSNs). Different from the conventional dKF, each adjacent node in the WSNs is detected to ascertain its trustworthiness before local estimate fusion, so as to form a new secure network topology. Then the combination step is performed to fuse the information collected from the secure topology. The proposed secure dKF algorithm, having a better estimation performance, is robust to false data injection attacks on multiple sensors and partial elements of measurements. For the proposed secure dKF algorithm, its mean and mean‐square performance are derived, based on which its convergence is analysed. Additionally, the estimating and tracking problem of projectile position is investigated to confirm the effectiveness of the proposed secure dKF algorithm. It is shown by simulations that the proposed secure dKF algorithm achieves a significant estimation performance gain.

In cooperative localization for autonomous underwater vehicles (AUVs), the practical stochastic noise may be heavy-tailed, and nonstationary distributed because of acoustic speed variation, multipath effect of acoustic channel, and changeable underwater environment. To address such noise, a novel heavy-tailed mixture (HTM) distribution is first proposed in this article, and then expressed as a hierarchical Gaussian form by employing a categorical distributed auxiliary vector. Based on that, a novel HTM distribution based robust Kalman filter is proposed, where the one-step prediction, and measurement likelihood probability density functions are, respectively, modeled as an HTM distribution, and a Normal-Gamma-inverse Wishart distribution. The proposed filter is verified by a lake experiment about cooperative localization for AUVs. Compared with the cutting-edge filter, the proposed filter has been improved by 50.27% in localization error but no more than twice computational time is required.

To better model the non-Gaussian heavy-tailed measurement noise with unknown and time-varying bias, a new Student's t-inverse-Wishart (STIW) distribution is presented. The STIW distribution is firstly written as a Gaussian, inverse-Wishart and normal-Gamma hierarchical form, from which a new robust Kalman filter is then derived based on the variational Bayesian method. Simulation results illustrate the potentials of the new derived robust Kalman filter for addressing the above measurement noise.

The selection of step sizes in the progressive Gaussian approximate filter (PGAF) is important, and it is difficult to select optimal values in practical applications. Furthermore, in the PGAF, significant integral approximation errors are generated by the repeated approximate calculations of the Gaussian weighted integrals, which results in an inaccurate measurement noise covariance matrix (MNCM). To solve these problems, in this paper, the step sizes and the MNCM are jointly estimated based on the variational Bayesian (VB) approach. By incorporating the adaptive estimates of step sizes and the MNCM into the PGAF framework, a novel PGAF with variable step size is proposed. Simulation results illustrate that the proposed filter has higher estimation accuracy than existing state-of-the-art nonlinear Gaussian approximate filters.

A novel variational Bayesian (VB)-based adaptive Kalman filter (AKF) is proposed to solve the filtering problem of a linear system with unknown probability of measurement loss. The sum of two likelihood functions is transformed into an exponential multiplication form, and the state vector, the Bernoulli random variable and the probability of measurement loss are jointly inferred based on the VB approach. Simulation results demonstrate the superiority of the proposed AKF as compared with the existing filtering algorithms with unknown probability of measurement loss.

In this paper, a new robust Kalman filtering framework for a linear system with non-Gaussian heavy-tailed and/or skewed state and measurement noises is proposed through modeling one-step prediction and likelihood probability density functions as Gaussian scale mixture (GSM) distributions. The state vector, mixing parameters, scale matrices, and shape parameters are simultaneously inferred utilizing standard variational Bayesian approach. As the implementations of the proposed method, several solutions corresponding to some special GSM distributions are derived. The proposed robust Kalman filters are tested in a manoeuvring target tracking example. Simulation results show that the proposed robust Kalman filters have a better estimation accuracy and smaller biases compared to the existing state-of-the-art Kalman filters.

In this paper, a novel robust Rauch-Tung-Striebel smoother is proposed based on the Slash and generalized hyperbolic skew Student's t-distributions. A novel hierarchical Gaussian state-space model is constructed by formulating the Slash distribution as a Gaussian scale mixture form and formulating the generalized hyperbolic skew Student's t-distribution as a Gaussian variance-mean mixture form, based on which the state trajectory, mixing parameters and unknown noise parameters are jointly inferred using the variational Bayesian approach. The posterior probability density functions of mixing parameters of the Slash and generalized hyperbolic skew Student's t-distributions are, respectively, approximated as truncated Gamma and generalized inverse Gaussian. Simulation results illustrate that the proposed robust Rauch-Tung-Striebel smoother has better estimation accuracy than existing state-of-the-art smoothers.

The scale matrix and degrees of freedom (dof) parameter of a Student's t distribution are important for nonlinear robust inference, and it is difficult to determine exact values in practical application due to complex environments. To solve this problem, an improved robust Gaussian approximate (GA) filter is derived based on the variational Bayesian approach, where the state together with unknown scale matrix and dof parameter are inferred. The proposed filter is applied to a target tracking problem with measurement outliers, and its performance is compared with an existing robust GA filter with fixed scale matrix and dof parameter. The results show the efficiency and superiority of the proposed filter as compared with the existing filter.

In this letter, a robust Gaussian approximate (GA) fixed-interval smoother for nonlinear systems with heavy-tailed process and measurement noises is proposed. The process and measurement noises are modeled as stationary Student's t distributions, and the state trajectory and noise parameters are inferred approximately based on the variational Bayesian (VB) approach. Simulation results show the efficiency and superiority of the proposed smoother as compared with existing smoothers.

A novel multimodal approach is proposed to solve the problem of blind source separation (BSS) of moving sources. The challenge of BSS for moving sources is that the mixing filters are time varying, thus the unmixing filters should also be time varying, which are difficult to track in real time. In the proposed approach, the visual modality is utilized to facilitate the separation for both stationary and moving sources. The movement of the sources is detected by a 3-D tracker based on particle filtering. The full BSS solution is formed by integrating a frequency domain blind source separation algorithm and beamforming: if the sources are identified as stationary, a frequency domain BSS algorithm is implemented with an initialization derived from the visual information. Once the sources are moving, a beamforming algorithm is used to perform real time speech enhancement and provide separation of the sources. Experimental results show that by utilizing the visual modality, the proposed algorithm can not only improve the performance of the BSS algorithm and mitigate the permutation problem for stationary sources, but also provide a good BSS performance for moving sources in a low reverberant environment.

A new variable step-size least-mean-square (VSSLMS) algorithm is presented in this paper for applications in which the desired response contains nonstationary noise with high variance. The step size of the proposed VSSLMS algorithm is controlled by the normalized square Euclidean norm of the averaged gradient vector, and is henceforth referred to as the NSVSSLMS algorithm. As shown by the analysis and simulation results, the proposed algorithm has both fast convergence rate and robustness to high-variance noise signals, and performs better than Greenburg's sum method, which is a robust algorithm for applications with nonstationary noise.

This letter proposes a new variable tap-length least-mean-square (LMS) algorithm for applications in which the unknown filter impulse response sequence has an exponential decay envelope. The algorithm is designed to minimize the mean-square deviation (MSD) between the optimal and adaptive filter weight vectors at each iteration. Simulation results show the proposed algorithm has a faster convergence rate as compared with the fixed tap-length LMS algorithm and is robust to the initial tap-length choice.

A new method, employing machine learning techniques and a modified low frequency envelope spectrum estimator, for estimating important room acoustic parameters including Reverberation Time (RT) and Early Decay Time (EDT) from received music signals has been developed. It overcomes drawbacks found in applying music signals directly to the envelope spectrum detector developed for the estimation of RT from speech signals. The octave band music signal is first separated into sub bands corresponding to notes on the equal temperament scale and the level of each note normalised before applying an envelope spectrum detector. A typical artificial neural network is then trained to map these envelope spectra onto RT or EDT. Significant improvements in estimation accuracy were found and further investigations confirmed that the non-stationary nature of music envelopes is a major technical challenge hindering accurate parameter extraction from music and the proposed method to some extent circumvents the difficulty.

A convex combination of adaptive filters is utilized to improve the performance of a variable tap-length least-mean-square (LMS) algorithm in a low signal-to-noise environment (SNRles0 dB). As shown by our simulations, the adaptation of the tap-length in the variable tap-length LMS algorithm is highly affected by the parameter choice and the noise level. Combination approaches can improve such adaptation by exploiting advantages of parallel adaptive filters with different parameters. Simulation results support the good properties of the proposed method