Xiaotong Tu

• Ph.D.
• Professor (Associate) at Xiamen University

Hyperspectral image (HSI) reconstruction aims to restore the original 3D HSIs from the 2D hyperspectral snapshot compressive images (SCIs). The key to high-fidelity HSI reconstruction lies in designing refined spatial and spectral attention mechanisms, which are crucial for generating fine-grained representations of HSI based on the limited spatial...

Pan-sharpening aims to preserve the spectral information of the multi-spectral (MS) image while leveraging the high-frequency details from the guided high-resolution panchromatic (PAN) image to enhance its spatial resolution. The key challenge is how to preserve the spectral information from the MS image and the spatial details from the PAN image a...

Low-light image enhancement (LIE) aims at precisely and efficiently recovering an image degraded in poor illumination environments. Recent advanced LIE techniques are using deep neural networks, which require lots of low-normal light image pairs, network parameters, and computational resources. As a result, their practicality is limited. In this wo...

Multivariate time series (MTS) forecasting aims to predict future patterns by extracting features from multi-variate history. Predominant methods face challenges in learning spatial dependencies while capturing long-term trends and local details, leading to suboptimal performance in MTS forecasting. To address this problem, we propose a Patch-Aware...

Anomaly detection (AD) in 3D point clouds is crucial in a wide range of industrial applications, especially in various forms of precision manufacturing. Considering the industrial demand for reliable 3D AD, several methods have been developed. However, most of these approaches typically require training separate models for each category, which is m...

Recent generalizable fault diagnosis researches have effectively tackled the distributional shift between unseen working conditions. Most of them mainly focus on learning domain-invariant representation through feature-level methods. However, the increasing numbers of unseen domains may lead to domain-invariant features contain instance-level spuri...

Underwater salient object detection (USOD) aims to identify the most crucial elements in underwater environments, holding significant potential for underwater exploration. Existing methods often overlook light degradation or involve larger network sizes, which are unsuitable for underwater mobile platforms and pose challenges to implement in practi...

Infrared and visible image fusion (IVIF) aims to effectively integrate complementary information from both infrared and visible modalities, enabling a more comprehensive understanding of the scene and improving downstream semantic tasks. Recent advancements in Mamba have shown remarkable performance in image fusion, owing to its linear complexity a...

Reconstruction-based methods, as one of the mainstream and advanced methods for anomaly detection, have attracted significant attention in the academic community. Although these methods may achieve good performance on some ideal industrial datasets, background factors have considerable influence on detecting anomalies due to a complex and ever-chan...

Low-light image enhancement (LIE) aims at precisely and efficiently recovering an image degraded in poor illumination environments. Recent advanced LIE techniques are using deep neural networks, which require lots of low-normal light image pairs, network parameters, and computational resources. As a result, their practicality is limited. In this wo...

The identification of nonlinear chirp signals has attracted notable attention in the recent literature, including estimators such as the variational mode decomposition and the nonlinear chirp mode estimator. However, most presented methods fail to process signals with close frequency intervals or depend on user-determined parameters that are often...

Industrial anomaly detection involves the identification and localization of abnormal regions in images, of which the core challenge is modeling normal data in appropriate ways. Inspired by dictionary learning, we propose a convolutional sparse reconstructive noise-robust framework, named AnoCSR. The proposed convolutional sparse encoding block (CS...

Structural health monitoring (SHM) is considered an effective approach to analyze the efficient working of several mechanical components. For this purpose, ultrasonic guided waves can cover long-distance and assess large infrastructures in just a single test using a small number of transducers. However, the working of the SHM mechanism can be affec...

Mechanical system condition monitoring is an important procedure in modern industry, which not only reduces maintenance costs but also ensures safe equipment operation. At present, the monitoring method based on signal processing is one of the most common and effective fault diagnosis methods. In this work, the time-frequency distribution (TFD) obt...

The decomposition of non-stationary signals remains a challenge in a wide variety of fields. Especially, the impulse or cross-mode signals are difficult to be reconstructed by recent methods due to their transient characteristic. Moreover, most methods rely heavily on the user-defined settings of the regularized parameter for the convex optimizatio...

We devise a new regularization for denoising with self-supervised learning. The regularization uses a deep image prior learned by the network, rather than a traditional predefined prior. Specifically, we treat the output of the network as a ``prior'' that we again denoise after ``re-noising.'' The network is updated to minimize the discrepancy betw...

Recently, many forms of audio industrial applications, such as sound monitoring and source localization, have begun exploiting smart multi-modal devices equipped with a microphone array. Regrettably, model-based methods are often difficult to employ for such devices due to their high computational complexity, as well as the difficulty of appropriat...

Siamese networks are widely used in various contrastive learning methods for recognition tasks, with few labeled data and abundant unlabeled data. In the field of fault diagnosis, it is universal to face the problem that large collections of common fault data and few catastrophic fault samples result in the imbalanced distribution of fault data col...

Nonlinear group delay signals with frequency-varying characteristics are common in a wide variety of fields, for instance, structural health monitoring and fault diagnosis. For such applications, the signal is composed of multiple modes, where each mode may overlap in the frequency-domain. The resulting decomposition and forming of time-frequency r...

Till now, most of the previous work has emphasized the functioning of several structural health monitoring (SHM) techniques and systematic changes in SHM execution. However, there exist insufficient data in the literature regarding the patent-based technological developments in the SHM research domain which might be a useful source of detailed info...

Underwater images suffer from degradation caused by light scattering and absorption. Training a deep neural network to restore underwater images is challenging due to the laborintensive data collection and the lack of paired data. To this end, we propose an unsupervised and untrained underwater image restoration method based on the layer disentangl...

In real-world scenarios, aerial image datasets are generally class imbalanced, where the majority classes have rich samples, while the minority classes only have a few samples. Such class imbalanced datasets bring great challenges to aerial scene recognition. In this paper, we explore a novel two-stage contrastive learning framework, which aims to...

Acoustic source localization has been applied in different fields, such as aeronautics and ocean science, generally using multiple microphones array data to reconstruct the source location. However, the model-based beamforming methods fail to achieve the high-resolution of conventional beamforming maps. Deep neural networks are also appropriate to...

Variational nonlinear chirp mode decomposition (VNCMD) is a recently introduced method for nonlinear chirp signal decomposition that has aroused notable attention in various fields. One limiting aspect of the method is that its performance relies heavily on the setting of the bandwidth parameter. To overcome this problem, we here propose a Bayesian...

Video-based action recognition is a challenging task, which demands carefully considering the temporal property of videos in addition to the appearance attributes. Particularly, the temporal domain of raw videos usually contains significantly more redundant or irrelevant information than still images. For that, this paper proposes an unsupervised v...

Time-frequency analysis (TFA) is a vital method to deal with nonstationary signals. However, the traditional TFA methods cannot produce a concentrated time-frequency representation (TFR) for the nonstationary signal with strongly time-varying instantaneous frequency (IF). In this work, a new TFA method named generalized synchrosqueezing transform i...

Beamforming technology plays a significant role in source localization and quantification. As traditional delay-andsum beamformers generally yield low spatial resolution, as well as suffer from the occurrence of spurious sources, different forms of deconvolution methods have been proposed in the literature. In this work, we propose two approaches b...

The instantaneous frequency (IF) is an important feature for the analysis of non-stationary signals. However, extracting multiple IF ridges, crossing IF ridges, and discontinuous IF ridges simultaneously is still a challenging task. To solve this problem, an adaptive IF ridge extraction method is proposed. This method regards the IF ridge values at...

The guided wave is an efficient and reliable tool for the structural health monitoring (SHM) of the composite laminates. In the guided wave-based SHM methods, extracting the dispersion curves is essential for integrity evaluation. In this study, a sparse wavenumber analysis based on hybrid least absolute shrinkage and selection operator (Lasso) reg...

Damage detection in a mechanical structure using ultrasonic guided waves becomes even more problematic when the effect of variation in environmental and operating conditions, such as mechanical noise, temperature, flow rate, inner pressure, etc. is taken into account. The variation in these environmental and operating conditions can degrade the acc...

Structural health monitoring (SHM) is recognized as an efficient tool to interpret the reliability of a wide variety of infrastructures. To identify the structural abnormality by utilizing the electromechanical coupling property of piezoelectric transducers, the electromechanical impedance (EMI) approach is preferred. However, in real-time SHM appl...

The decomposition of nonlinear chirp modes is a challenging task, typically requiring prior knowledge of the number of modes a signal contains. In this work, we present a greedy nonlinear chirp mode estimation (NCME) technique that forms the used decomposition basis from the signal itself, using an arctangent demodulation technique. The resulting d...

Synchrosqueezing transform (SST) is a currently proposed novel post-processing time-frequency (TF) analysis tool. It has been widely shown that SST is able to improve the TF representation. However, so far how to improve the TF resolution while ensuring the accuracy of signal reconstruction is still an open question, particularly for the vibration...

The instantaneous speed (IS), i.e. instantaneous frequency, is an important feature to exploit the time-frequency characteristic of nonstationary signals. The IS estimation is widely used in various fields of signal processing. The IS estimation can also be regarded as a useful tool for rotating machinery fault diagnosis under nonstationary conditi...

Time–frequency analysis is recognized as an efficient tool to characterize the time-varying feature from the oscillatory signal by transforming it into an identifiable form. Some traditional time–frequency transforms are subjected to poor time–frequency resolution or do not allow for mode reconstruction. As a postprocessing method, the synchrosquee...

This paper proposes an adaptive short-time Fourier transform (ASTFT) method based on fast path optimization (FPO). The proposed method is applied to the fault diagnosis of planetary gearbox. This method can estimate the instantaneous frequency from the time-frequency distribution. Thus, the time window size can be adaptively adjusted by considering...

Timefrequency analysis (TFA) is regarded as an efficient technique to reveal the hidden characteristics of the oscillatory signal. At present, the traditional TFA methods always construct the signal model in the time domain and assume the instantaneous features of the modes to be continuous. Thus, most of these approaches fail to tackle some speci...

2019 IEEE TIM Outstanding Reviewer.

Advancements in the synchrosqueezing transform as a postprocessing time–frequency method have received considerable attention in the past few decades for the analysis of nonstationary signals. Many studies have focused on improving the accuracy of the estimated instantaneous frequency (IF). In some fields, the IF spectra of signals exhibit fast var...

This paper introduces the second-order transient-extracting transform (TET2) to extract transient components from a nonstationary signal. Different from the traditional transient-extracting transform (TET1), the proposed method is based on a more general frequency-domain signal model, termed Gaussian-modulated linear group delay signal model. The f...

Empirical Wavelet Transform (EWT) has shown its effectiveness in some applications. However, when noisy and non-stationary signals are analyzed, some local maxima may appear and be retained in the peak sequence mistakenly, an improper segmentation in the frequency domain will occur. In our research, morphological empirical wavelet transform (MEWT)...

The Lamb wave inspection has emerged as a promising method for structural health monitoring and nondestructive testing. However, because of the highly dispersive and multimodal features, the Lamb wave mode separation has become a challenging problem. Based on the dispersion curve analysis, a new signal processing method is proposed in this study to...

This paper considers the analysis of impulsive-like signals whose time-frequency ridge curves are nearly perpendicular to the time axis. Although the instantaneous frequency of such a signal is a multivalued time-dependent function, its group delay is a single-valued function of frequency, which indicates that a frequency-domain signal model is mor...

Nowadays, each industry pays much attention to condition monitoring to ensure operational safety. The vibration-based method is regarded as an effective technique for mechanical conditional monitoring. Due to some machines usually operating under a harsh condition, it causes a time-varying instantaneous frequency (IF) feature in the vibration signa...

Matching demodulation is a new method of time-frequency analysis. The paper used matching demodulation and synchrosqueezing technology to extract ridges of frequency from multi-component signal spectrum. The ridges of weak energy component signals are obtained by using translation operator filtering and envelope line filtering. By using this method...

Bearings are widely used in rotating machinery such as aircraft engines and wind turbines. In this paper, we proposed a new data-driven method called frozen convolution and activated memory network (FCAMN) for bearing remaining useful life (RUL) estimation based on the deep neural network. The proposed method is composed of two parts: the multi-sca...

The time-frequency analysis (TFA) is a robust technique for instantaneous frequency (IF) estimation and component separation of nonstationary signal. In this paper, a new method called parameterized synchrosqueezing transform (PST) is proposed. The PST introduces a two-step algorithm, including parameters estimation and synchrosqueezing method to a...

The synchrosqueezing transform (SST) is a powerful tool for time-frequency analysis of signals with slowly varying instantaneous frequency (IF). However, the SST and its extensions provide poor time-frequency resolution for signals with wide frequency range and fast varying IF. In this paper, a new SST method called high-order synchrosqueezing wave...

As one of the most important and essential elements of machines, rolling element bearings always fail due to the severe operation environment. Bearing failures usually result in the periodic impulses, which are the most crucial feature for the bearing diagnosis. These impulses may be overwhelmed by the background noises or other unrelated component...

The rolling element bearing is easy to be malfunctioning due to the harsh operation. When a fault exists in the bearing, it can generate the periodical or quasi-periodical impulses, which are important features for the bearing fault detection. These impulses may be submerged in the background noise and interferences of other unrelated components. T...

Vibration monitoring is an effective method for mechanical fault diagnosis. Wind turbines usually operated under varying-speed condition. Time-frequency analysis (TFA) is a reliable technique to handle such kind of nonstationary signal. In this paper, a new scheme, called current-aided TFA, is proposed to diagnose the planetary gearbox. This new te...

The planetary gearbox is one of the key components in the rotating machinery. The planetary gearbox is prone to malfunction, which increases downtime and repair costs. Hence, the fault diagnosis of the planetary gearbox is an important research topic. The acquired signal from the planetary gearbox exhibit strongly time-variant and nonstationary fea...

Time-frequency analysis (TFA) is considered as a useful tool to extract the time variant features of the nonstationary signal. In this paper, a new method called demodulated high-order synchrosqueezing transform (DHST) is proposed. The DHST introduces a two-step algorithm, namely demodulated transform (DT) and high-order synchrosqueezing method to...

Wind turbines usually operate under nonstationary conditions, such as wide-range speed fluctuation and time-varying load. Its critical component, the planetary gearbox, is prone to malfunction or failure, which leads to downtime and repair costs. Therefore, fault diagnosis and condition monitoring for the planetary gearbox in wind turbines is a vit...

Most of researcher assume that there exist cubic nonlinear term directly when investigate the non-linear phenomena of torsional vibration of rotor. However, they do not explore the reason why emerge non-linear phenomena. In this paper, softing duffing non-linear terms is gotten in torsional vibration of rotor with geometry deformation. Then modal a...

The Empirical Wavelet Transform (EWT) is a novel method for analyzing the multi-component signals by constructing an adaptive filter bank. Though it is an effective tool to identify the signal components, it has it has drawback in dealing with some noisy and non-stationary signals due to its coarse spectrum segmentation. To target this problem, an...

This paper presents a novel approach to detect the milling chatter based on energy entropy. By using variational mode decomposition and wavelet packet decomposition, the cutting force signal is decomposed into two group of sub-signals respectively, and each component has limited bandwidth in spectral domain. Since milling chatter is characterized b...