ZWZhaoyue Wu

Zhaoyue Wu

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Zhaoyue Wu has a total of 41 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Monitoring Vegetation Structure, Function and Composition with Remote Sensing

    by Fabian D. Schneider, Ting Zheng, Antonio Ferraz, Laura Berman, Camilla D. Jakobsen, Jaime C. Revenga, Zhaoyue Wu, Zhiwei Ye, Ryan P. Pavlick, Philip A. Townsend, et al.

    Biodiversity monitoring is important to support decision-making for managing landscapes sustainably and supporting national and international policy targets for nature conservation and restoration, including the Global Biodiversity Framework. While assessing the status, change and drivers of biodiversity remains challenging, we have new opportunities to support biodiversity monitoring from space with a growing suite of Earth observation satellites. Remote sensing is especially well suited to monitor ecosystems in terms of their vegetation structure and forest structural diversity with lidar and radar, as well as vegetation functions, foliar functional diversity and community composition with imaging spectroscopy and multispectral imaging. In this talk, we will provide examples for monitoring forest structural diversity using the spaceborne lidar GEDI. We evaluated and compared structural diversity across contrasting biomes in the Western US and Central Africa, and we found that general biogeographic patterns of higher horizontal structural diversity in areas with higher disturbance, higher topographic variation and lower aridity hold across continents and scales. For monitoring ecosystem functions, we will provide examples for monitoring plant traits and functional diversity using imaging spectroscopy along elevation gradients in California. We will provide insights into the role of trait-trait relationships and trait selection for mapping trait diversity patterns at the landscape scale. We found that diversity patterns vary by the type and number of functional traits included in the analyses, and that the interpretation is context dependent. And for monitoring composition, we will provide examples indicating how well we can distinguish different vegetation types, communities and species with spectroscopy, and how well we predict animal composition and niche space using a remote sensing-based biodiversity data cube, BioCube. With these examples, we will demonstrate new capabilities and avenues for monitoring different aspects of biodiversity change using remote sensing at the landscape scale, and we will provide important context for the interpretation of these results. Remote sensing can provide information about biological communities and habitats, ecosystems and biomes at different spatial scales and time steps that should be integrated with other biodiversity data, models and decision support tools to fully leverage its potential for biodiversity monitoring.

  • Open Access

    Monitoring Ecosystem Structure, Function and Composition with Remote Sensing

    by Fabian D. Schneider, Ting Zheng, Antonio Ferraz, Laura Berman, Camilla D. Jakobsen, Jaime C. Revenga, Zhaoyue Wu, Zhiwei Ye, Ryan P. Pavlick, Philip A. Townsend, et al.

    Biodiversity monitoring is important to support decision-making for managing landscapes sustainably and supporting national and international policy targets for nature conservation and restoration, including the Global Biodiversity Framework. While assessing the status, change and drivers of biodiversity remains challenging, we have new opportunities to support biodiversity monitoring from space with a growing suite of Earth observation satellites. Remote sensing is especially well suited to monitor ecosystems in terms of their vegetation structure and forest structural diversity with lidar and radar, as well as vegetation functions, foliar functional diversity and community composition with imaging spectroscopy and multispectral imaging. In this talk, we will provide examples for monitoring forest structural diversity using the spaceborne lidar GEDI. We evaluated and compared structural diversity across contrasting biomes in the Western US and Central Africa, and we found that general biogeographic patterns of higher horizontal structural diversity in areas with higher disturbance, higher topographic variation and lower aridity hold across continents and scales. For monitoring ecosystem functions, we will provide examples for monitoring plant traits and functional diversity using imaging spectroscopy along elevation gradients in California. We will provide insights into the role of trait-trait relationships and trait selection for mapping trait diversity patterns at the landscape scale. We found that diversity patterns vary by the type and number of functional traits included in the analyses, and that the interpretation is context dependent. And for monitoring composition, we will provide examples indicating how well we can distinguish different vegetation types, communities and species with spectroscopy, and how well we predict animal composition and niche space using a remote sensing-based biodiversity data cube, BioCube. With these examples, we will demonstrate new capabilities and avenues for monitoring different aspects of biodiversity change using remote sensing at the landscape scale, and we will provide important context for the interpretation of these results. Remote sensing can provide information about biological communities and habitats, ecosystems and biomes at different spatial scales and time steps that should be integrated with other biodiversity data, models and decision support tools to fully leverage its potential for biodiversity monitoring.

  • Open Access

    From Multispectral to Hyperspectral: Cross-Sensor Intercomparison of Spectral Characteristics and Diversity Estimation in the Åmose Wetland Restoration Landscape

    by Zhaoyue Wu, Signe Normand, Andreas Hueni, Marius Vögtli, Fabian Schneider

    Imaging spectroscopy has become an essential tool for biodiversity monitoring, enabling the characterization of ecosystem composition, function, and diversity through spectral information, grounded in the spectral variation hypothesis. However, the growing diversity of sensors and platforms introduces both new opportunities and challenges for consistent biodiversity assessment from local to global scales. Differences in spectral resolution, spatial resolution, and signal-to-noise ratio can substantially affect the measured spectral response of the earth’s surface and, consequently, the estimation of spectral diversity. To address these issues, this study conducts a systematic cross-sensor comparison focusing on emerging spaceborne hyperspectral systems (PRISMA and EnMAP), the spaceborne multispectral system (Sentinel-2), and the airborne imaging spectrometer (AVIRIS-4), aiming to evaluate differences in spectral characteristics and diversity estimation for long-term and large-scale biodiversity monitoring. Specifically, a multi-source reflectance dataset was acquired over the Åmose wetland restoration landscape in Denmark in August 2025 and processed with BRDF and geometric co-registration corrections to eliminate undesired view-sun-angle and geometric differences between sensors. Then, spectral intercomparison was performed both for images before and after BRDF correction under unified spatial and spectral resolution by extracting reflectance spectra from several representative land use classes (including nature, intensive agriculture, extensive agriculture, coniferous forest, deciduous forest, water, and built-up) to identify reliable wavelength ranges for subsequent spectral diversity estimation. For the spectral diversity estimation, the high-resolution AVIRIS-4 data (1 m) were resampled to multiple coarser resolutions (e.g. 10, 20, 30, and 60 m). Spectral diversity metrics were computed at each resolution and compared with corresponding spaceborne sensors (PRISMA and EnMAP at 30 m and Sentinel-2 at 10 m and 20 m), revealing how spatial scale and sensor characteristics influence the estimation of spectral diversity. In addition, uncertainty analysis was employed to quantify the variability and assess the reliability of both spectral intercomparison and diversity estimation. In general, this study provides a framework for clarifying the differences in spectral response and spatial resolution of promising sensors for spectral diversity estimation to support consistent biodiversity monitoring. The outcomes contribute to the development of standardized approaches for cross-sensor spectral diversity estimation and uncertainty reduction in next-generation remote sensing of biodiversity.

  • Open Access

    A Physics-Guided Diffusion Unmixing Model for Global Lunar Mineral Abundance Mapping

    by Pan Zheng, Hongjun Su, Zhaoyue Wu, Juan M. Haut, Mercedes E. Paoletti

    Accurate global mapping of lunar regolith minerals is critical for understanding the geological evolution and resource potential of the Moon. However, spectral unmixing algorithms originally designed for terrestrial remote sensing often perform poorly when applied to the Moon, due to the nonlinear intimate mixing and spectral variability associated with its unique surface environment. To address these challenges, this study proposes a novel physics-guided radiative transfer model with a deep diffusion network, informed by spectral library prior, to estimate lunar mineral abundances from Chang'E-1 Interferometric Imaging Spectrometer (IIM) hyperspectral data. Single-scattering albedo (SSA) is retrieved by inverting the Hapke model, with photometric normalization to deal with multiple-scattering effects, observation geometry, and macroscopic roughness to mitigate extrinsic effects prior to unmixing. A physics-guided spectral library is constructed to capture intrinsic endmember variability arising from mineralogy and grain size. Spectral unmixing is then performed in the SSA domain using a diffusion model coupled with Bayesian abundance estimation. Validation against lunar sample data and 19 verified lunar sites confirms the accuracy of the proposed method. Global abundance maps for plagioclase, glass, orthopyroxene, olivine, ilmenite, and clinopyroxene exhibit improved spatial detail and better geological coherence compared to previous methods. These results validate the proposed method as a robust approach for global-scale lunar mineral mapping, with strong potential for integration into future planetary remote sensing missions.


  • DenseMixerMamba: Residual Mixing for Spectral–Spatial Hyperspectral Image Classification

    by Mercedes E. Paoletti, Zhaoyue Wu, Pan Zheng, Danfeng Hong, Juan M. Haut

    The classification of hyperspectral images (HSIs) is a fundamental but challenging topic in the field of surface monitoring. In recent years, deep learning models like RNN, CNN, Transformers, and Mamba have demonstrated their remarkably and promising processing capabilities to confront high-dimensional and complicated HSI data. However, how to capture the refined spatial and spectral information from high-dimensional data continues to pose efficiency problems, which are compounded by the increased complexity of deep models. CNNs and Transformers employ thousands of parameters, imposing a high computational load and memory consumption. This paper addressed the parameter reduction problem by proposing an end-to-end DenseMixerMamba network for HSI classification, which converges Mamba, attention mechanisms, and dense architectures. Specifically, channel attention is first introduced to capture valuable channels and stimulate subsequent feature representation quickly. Then, to enhance the flow of information between layers and provide more stable training, a novel encoder based on the Mamba Vision Mixer (MVM) block and dense connectivity is developed to capture spatial and spectral patterns without increasing computational effort. Furthermore, a dual-branch architecture is designed to support the MVM block, which maintains both local and global information for final representation. Numerous experiments on four real datasets exhibit the more impressive classification accuracy and generalization performance of DenseMixerMamba over other state-of-the-art methods.


  • Open Access

    Graph Kolmogorov–Arnold Convolutional Network for Hyperspectral Image Classification

    by Hengyi Zheng, Hongjun Su, Zhaoyue Wu, Yiping Chen, Qian Du

    The Graph Neural Network (GNN) exhibits noteworthy performance in hyperspectral image classification (HSIC) due to its efficient message passing structure, which employs the Multi-Layer Perceptron (MLP) for feature extraction and utilizes nonlinear activation function between two layers of the network. However, two significant limitations (poor model scaling and lack of interpretability) impede GNN's ability to extract deeper features in the graph. Therefore, this paper proposes a novel method called Graph Kolmogorov-Arnold Convolutional Network (GKACN) where the Kolmogorov-Arnold Network (KAN) theory based convolutional operation employed in the graph for HSIC. As a new network architecture designed to replace the traditional MLP, KAN addresses the model scaling and interpretability limitations of MLP by using learnable spline-based functions in lieu of weights, resulting GKACN fully exploits the potential of the KAN and GNN structures to facilitate the mining of more representative features. Moreover, textural features from the extended morphological attribute profile extraction are used for spatial-spectral classification. Comprehensive experiments on four distinct HSI datasets have demonstrated the feasibility and superiority of the proposed method compared to other state-of-the-art models.


  • KACNet: Kolmogorov-Arnold Convolution Network for Hyperspectral Anomaly Detection

    by Zhaoyue Wu, Hailiang Lu, Mercedes E. Paoletti, Hongjun Su, Weipeng Jing, Juan M. Haut

    Hyperspectral images capture numerous narrow spectral bands to provide detailed information to identify and locate targets, making them highly suitable for anomaly detection tasks. In recent years, deep learning techniques have demonstrated impressive capabilities and prospects in hyperspectral anomaly detection, primarily relying on multilayer perceptrons (MLPs) and convolutional neural networks (CNNs) to extract and represent nonlinear features. However, MLPs and CNNs often require deeper network architectures when dealing with complex high-dimensional data, resulting in a constrained generalization and limited representation of features. To address this issue, and inspired by the recent Kolmogorov-Arnold Network (KAN), this paper introduces a novel asymmetric convolutional autoencoder network by integrating KAN and CNN, named KACNet. Specifically, we design a spectral KAN block in the convolutional encoder and a spatial KAN block in the convolutional decoder, to simultaneously enhance the feature extraction and characterization capabilities of the network. Furthermore, to effectively utilize the limited prior information, a weight initialization mechanism based on Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) is developed to boost the background recovery. By combining KAN, CNN, and HDBSCAN, the proposed integration enhances the interpretability and reliability of hyperspectral anomaly detection. Extensive experiments are conducted on six public datasets, demonstrating that the KAN poses remarkable performance on background reconstruction, particularly, the proposed KACNet significantly outperforms the other state-of-the-art methods.


  • Weighted Spatiotemporal Fusion via Tensor Collaborative Representation

    by Hui Shen, Hongjun Su, Hongliang Lu, Zhaoyue Wu, Qian Du

    Spatio-temporal fusion of remote sensing data is one of the critical techniques for Earth’s surface dynamic monitoring and analysis, which solves the limitation of spatial resolution and temporal coverage in individual sensor. In order to establish a more accurate and physically meaningful spatio-temporal fusion model, a weighted spatio-temporal fusion method via tensor collaborative representation (W-STFTCR) is proposed. Specifically, the collaborative representation (CR) constraint is incorporated into the tensor decomposition framework to prevent overfitting and enhance model robustness. Meanwhile, the superpixel segmentation strategy is adopted to partition the input difference image into superpixel blocks, facilitating block dictionary construction and clustering effectively. In addition, the normalized difference vegetation index (NDVI) and joint information entropy are introduced for weighting bands in predicting the final image, which leads to more accurate and physically meaningful outcomes. To verify the performance of the proposed method, the spatio-temporal fusion experiments on two publicly available datasets were conducted. The experiment results show that the proposed method outperforms the previous state-of-the-art (SOTA) spatio-temporal fusion algorithms, with excellent parameter robustness.


  • SMCNet: Sparse-Inspired Masked Convolutional Network for Hyperspectral Anomaly Detection

    by Zhaoyue Wu, Mercedes E. Paoletti, Hongjun Su, Pan Zheng, Zhong Zhang, Juan M. Haut, Antonio Plaza

    Hyperspectral anomaly detection, which aims to blindly search and localize potential targets, is a research area with extensive application prospects and profound implications. In recent years, the emergence of unsupervised and self-supervised deep learning for image reconstruction has provided inspiring solutions for hyperspectral anomaly detection. However, owing to sensor-induced and environmental effects, the full-image detection networks inevitably reconstruct anomalies along with the background. Existing detectors indirectly mitigate anomaly reconstruction by imposing constraints on hidden features or loss functions, but they provide unsatisfactory performance in large target detection scenarios. This work straightforwardly addresses this issue from the input source, i.e., introducing the concept of masked autoencoders into fully convolutional networks and further developing a sparse-inspired masked convolutional network (SMCNet) consisting of three mutually supportive components: 1) a hierarchical encoder, 2) a sparse projection layer, and 3) a hierarchical decoder. The encoder employs an adaptive potential anomaly masking strategy, leveraging sparse convolution for extracting multidimensional features of the remaining background. Meanwhile, a sparse-guided projection layer is created by discarding the positional embedding technique to populate the uncoded region and guide the background recovery without introducing anomalies. Finally, the decoder couples the hierarchical structure and a hybrid attention mechanism (local-middle-global and spatial-spectral) to refine the background during image recovery, whereas anomalies in the residual map are highlighted. Extensive experiments using ten typical competitors on six different types of datasets validate the effectiveness and generalization ability of the newly proposed SMCNet method.