Tao Chen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Tao Chen has a total of 16 co-authors. Here are the top co-authors by number of shared publications:
- Xiaoxi Hu
School of Mathematics and Computer Science, Shaanxi University of Technology, Hanzhong, China
5 shared publications - Guangyi Ji
School of Mathematics and Computer Science, Shaanxi University of Technology, Hanzhong, China
4 shared publications - Jiaxi Hu
School of Mathematics and Computer Science, Shaanxi University of Technology, Hanzhong 723001, China
3 shared publications - Longquan Yong
School of Mathematics and Computer Science, Shaanxi University of Technology, Hanzhong, China
3 shared publications - Qian Chen
School of Mathematics and Computer Science, Shaanxi University of Technology, Hanzhong 723001, China
3 shared publications - Yuanhong Zhou
School of Mathematics and Computer Science, Shaanxi University of Technology, Hanzhong 723001, China
3 shared publications
Latest Publications
- 2 November 2026
- Published by Elsevier in Neurocomputing
- Vol. 712
SegFlux: Physical morphology flow and analytical routing for volumetric medical image segmentation
- 2 November 2026
- Published by Elsevier in Neurocomputing
- Vol. 712
SegFlux: Physical morphology flow and analytical routing for volumetric medical image segmentation
- 26 September 2026
- Published by Elsevier in Biomedical Signal Processing and Control
- Vol. 130
AC-UKAN: An enhanced U-KAN framework integrating modulated deformable convolution and mixed pooling
- 26 September 2026
- Published by Elsevier in Biomedical Signal Processing and Control
- Vol. 130
AC-UKAN: An enhanced U-KAN framework integrating modulated deformable convolution and mixed pooling
- 7 September 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 352
SSKD-Depth: Spectral–semantic Knowledge Distillation for efficient monocular depth estimation
- 7 September 2026
- Published by Elsevier in Knowledge-Based Systems
- Vol. 352
SSKD-Depth: Spectral–semantic Knowledge Distillation for efficient monocular depth estimation
- Open Access
- 3 July 2026
- Published by MDPI AG in Journal of Imaging
- Vol. 12 (7) , 299
MSCF-Net: A Vision Mamba Network with Multi-Scale Context Bridging and Cross-Layer Adaptive Fusion for Medical Image Segmentation
Accurate medical image segmentation remains challenging when lesions have large-scale variation, weak boundaries, and strong background interference. Vision Mamba provides efficient long-range modeling, but current Mamba-based U-shaped networks are still limited by weak local multi-scale representation and coarse skip fusion. This study proposes MSCF-Net, a Vision Mamba segmentation network for dermoscopic and endoscopic images. The network is built on VM-UNet and introduces two modules. The Multi-Scale Context Bridging (MSCB) module enriches bottleneck features with local, dilated, and global context. The Cross-Layer Adaptive Fusion (CLAF) module recalibrates encoder–decoder features in channel and spatial dimensions, reducing noisy shallow feature transmission. A structure loss is used to improve region completeness and boundary quality. Experiments on ISIC 2017, ISIC 2018, and CVC-ClinicDB show Dice scores of 90.62%, 90.82%, and 91.72%, and mIoU values of 82.02%, 82.31%, and 84.56%, respectively. Compared with representative baselines evaluated in our experiments, MSCF-Net achieves competitive segmentation performance under the adopted benchmark protocol. Ablation, qualitative, and spatial response analyses further indicate that MSCB improves scale-aware representation, while CLAF helps the decoder focus on lesion-related cues. The results suggest that MSCF-Net provides a favorable accuracy–efficiency trade-off for medical image segmentation.
- 3 July 2026
- Published by MDPI AG in Journal of Imaging
- Vol. 12 (7) , 299
Open AccessMSCF-Net: A Vision Mamba Network with Multi-Scale Context Bridging and Cross-Layer Adaptive Fusion for Medical Image Segmentation
Accurate medical image segmentation remains challenging when lesions have large-scale variation, weak boundaries, and strong background interference. Vision Mamba provides efficient long-range modeling, but current Mamba-based U-shaped networks are still limited by weak local multi-scale representation and coarse skip fusion. This study proposes MSCF-Net, a Vision Mamba segmentation network for dermoscopic and endoscopic images. The network is built on VM-UNet and introduces two modules. The Multi-Scale Context Bridging (MSCB) module enriches bottleneck features with local, dilated, and global context. The Cross-Layer Adaptive Fusion (CLAF) module recalibrates encoder–decoder features in channel and spatial dimensions, reducing noisy shallow feature transmission. A structure loss is used to improve region completeness and boundary quality. Experiments on ISIC 2017, ISIC 2018, and CVC-ClinicDB show Dice scores of 90.62%, 90.82%, and 91.72%, and mIoU values of 82.02%, 82.31%, and 84.56%, respectively. Compared with representative baselines evaluated in our experiments, MSCF-Net achieves competitive segmentation performance under the adopted benchmark protocol. Ablation, qualitative, and spatial response analyses further indicate that MSCB improves scale-aware representation, while CLAF helps the decoder focus on lesion-related cues. The results suggest that MSCF-Net provides a favorable accuracy–efficiency trade-off for medical image segmentation.
- Open Access
- 11 February 2026
- Published by MDPI AG in Preprints.org
Some New Results on N(2,2,0)-Algebras
An N(2,2,0)-algebra (abbreviated as NA-algebra) is an algebraic structure equipped with two binary operations, $\ast$ and $\bigtriangleup$, satisfying specific axioms. This paper investigates a special class of NA-algebras where the operation "$\ast $" exhibits nilpotent properties. We study several fundamental concepts within NA-algebras, including ideals, congruence decomposition, congruence kernels, and multiplicative stabilizers. A notion of NA-morphism is introduced, and a corresponding NA-morphism theorem is established. Furthermore, we explore the relationships between NA-algebras and other related logical algebraic structures, such as quantum B-algebras, Q-algebras, CI-algebras, pseudo-BCH-algebras, and RM-algebras. Notably, we prove that any nilpotent NA-algebra forms a quantum B-algebra. These results lay a foundation for further research into the structure and potential applications of NA-algebras.- 11 February 2026
- Published by MDPI AG in Preprints.org
Open AccessSome New Results on N(2,2,0)-Algebras
An N(2,2,0)-algebra (abbreviated as NA-algebra) is an algebraic structure equipped with two binary operations, $\ast$ and $\bigtriangleup$, satisfying specific axioms. This paper investigates a special class of NA-algebras where the operation "$\ast $" exhibits nilpotent properties. We study several fundamental concepts within NA-algebras, including ideals, congruence decomposition, congruence kernels, and multiplicative stabilizers. A notion of NA-morphism is introduced, and a corresponding NA-morphism theorem is established. Furthermore, we explore the relationships between NA-algebras and other related logical algebraic structures, such as quantum B-algebras, Q-algebras, CI-algebras, pseudo-BCH-algebras, and RM-algebras. Notably, we prove that any nilpotent NA-algebra forms a quantum B-algebra. These results lay a foundation for further research into the structure and potential applications of NA-algebras. - 3 November 2025
- Published by Elsevier in Knowledge-Based Systems
- Vol. 332
MonoPAM: Roadside monocular 3D object detection with polygonal attention mechanism
- 3 November 2025
- Published by Elsevier in Knowledge-Based Systems
- Vol. 332
MonoPAM: Roadside monocular 3D object detection with polygonal attention mechanism
- Open Access
- 11 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. PP, 1
HiFused-Depth: High-Frequency Fused Depth Estimation with Knowledge Guidance
Monocular depth estimation (MDE) is a fundamental task in computer vision with a wide range of applications. However, achieving accurate depth prediction remains challenging in scenes with complex structures and low texture areas. To address this issue, we propose a new depth estimation network that integrates high-frequency information enhancement and knowledge extraction. Specifically, the Sobel operator is used to extract high-frequency features such as edges and textures from the input RGB image, and merge them into the feature representation to enrich geometric understanding. Establish a high-frequency perceptual loss function to guide the network to focus on texture defects and structurally blurry areas, improving depth prediction accuracy. In addition, adopting a knowledge distillation strategy from a pretrained external teacher network, which transfers both global semantic context and fine-grained geometric cues to the student model, enhances generalization ability in different environments. Extensive experiments conducted on the NYUv2 and KITTI datasets have shown that the proposed method consistently outperforms existing methods in multiple standard evaluation metrics, achieving up to 7.3% reduction in log10 error and 6.3% reduction in Abs Rel on NYUv2, and 6.9% reduction in Sq Rel on KITTI compared with the strong NeWCRFs baseline.
- 11 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. PP, 1
Open AccessHiFused-Depth: High-Frequency Fused Depth Estimation with Knowledge Guidance
Monocular depth estimation (MDE) is a fundamental task in computer vision with a wide range of applications. However, achieving accurate depth prediction remains challenging in scenes with complex structures and low texture areas. To address this issue, we propose a new depth estimation network that integrates high-frequency information enhancement and knowledge extraction. Specifically, the Sobel operator is used to extract high-frequency features such as edges and textures from the input RGB image, and merge them into the feature representation to enrich geometric understanding. Establish a high-frequency perceptual loss function to guide the network to focus on texture defects and structurally blurry areas, improving depth prediction accuracy. In addition, adopting a knowledge distillation strategy from a pretrained external teacher network, which transfers both global semantic context and fine-grained geometric cues to the student model, enhances generalization ability in different environments. Extensive experiments conducted on the NYUv2 and KITTI datasets have shown that the proposed method consistently outperforms existing methods in multiple standard evaluation metrics, achieving up to 7.3% reduction in log10 error and 6.3% reduction in Abs Rel on NYUv2, and 6.9% reduction in Sq Rel on KITTI compared with the strong NeWCRFs baseline.
- Open Access
MonoAMP: Adaptive Multi-Order Perceptual Aggregation for Monocular 3D Vehicle Detection
Monocular 3D object detection is rapidly emerging as a key research direction in autonomous driving, owing to its resource efficiency and ease of implementation. However, existing methods face certain limitations in cross-dimensional feature attention mechanisms and multi-order contextual information modeling, which constrain their detection performance in complex scenes. Thus, we propose MonoAMP, an adaptive multi-order perceptual aggregation algorithm for monocular 3D object detection. We first introduce triplet attention to enhance the interaction of cross-dimensional feature attention. Second, we design an adaptive multi-order perceptual aggregation module. It dynamically captures multi-order contextual information and employs an adaptive aggregation strategy to enhance target perception. Finally, we propose an uncertainty-guided adaptive depth ensemble strategy, which models the uncertainty distribution in depth estimation and effectively fuses multiple depth predictions. Experiments demonstrate that MonoAMP significantly enhances performance on the KITTI dataset at the moderate difficulty level, achieving 16.80% and 24.47% . Additionally, the ablation study shows a 3.78% improvement in object detection accuracy over the baseline method. Compared to other advanced methods, MonoAMP demonstrates superior detection capabilities, especially in complex scenarios.Open AccessMonoAMP: Adaptive Multi-Order Perceptual Aggregation for Monocular 3D Vehicle Detection
Monocular 3D object detection is rapidly emerging as a key research direction in autonomous driving, owing to its resource efficiency and ease of implementation. However, existing methods face certain limitations in cross-dimensional feature attention mechanisms and multi-order contextual information modeling, which constrain their detection performance in complex scenes. Thus, we propose MonoAMP, an adaptive multi-order perceptual aggregation algorithm for monocular 3D object detection. We first introduce triplet attention to enhance the interaction of cross-dimensional feature attention. Second, we design an adaptive multi-order perceptual aggregation module. It dynamically captures multi-order contextual information and employs an adaptive aggregation strategy to enhance target perception. Finally, we propose an uncertainty-guided adaptive depth ensemble strategy, which models the uncertainty distribution in depth estimation and effectively fuses multiple depth predictions. Experiments demonstrate that MonoAMP significantly enhances performance on the KITTI dataset at the moderate difficulty level, achieving 16.80% and 24.47% . Additionally, the ablation study shows a 3.78% improvement in object detection accuracy over the baseline method. Compared to other advanced methods, MonoAMP demonstrates superior detection capabilities, especially in complex scenarios. - Open Access
Explicit Runge–Kutta Numerical Manifold Method for Solving the Burgers’ Equation via the Hopf–Cole Transformation
This paper presents an efficient numerical manifold method for solving the Burgers’ equation. To improve accuracy and streamline the solution process, we apply a nonlinear function transformation technique that reformulates the original problem into a linear diffusion equation. We utilize a dual cover mesh along with an explicit multi-step time integration method for spatial and temporal discretization, respectively. Constant cover functions are employed across mathematical covers, interconnected by a linear weight function for each manifold element. The full discretization formulation is derived using the Galerkin weak form. To efficiently compute the inverse of the symmetric positive definite mass matrix, we employ the Crout algorithm. The performance and convergence of our method are rigorously evaluated through several benchmark numerical tests. Extensive comparisons with exact solutions and alternative methods demonstrate that our approach delivers an accurate, stable, and efficient computational scheme for the Burgers’ equation.
Open AccessExplicit Runge–Kutta Numerical Manifold Method for Solving the Burgers’ Equation via the Hopf–Cole Transformation
This paper presents an efficient numerical manifold method for solving the Burgers’ equation. To improve accuracy and streamline the solution process, we apply a nonlinear function transformation technique that reformulates the original problem into a linear diffusion equation. We utilize a dual cover mesh along with an explicit multi-step time integration method for spatial and temporal discretization, respectively. Constant cover functions are employed across mathematical covers, interconnected by a linear weight function for each manifold element. The full discretization formulation is derived using the Galerkin weak form. To efficiently compute the inverse of the symmetric positive definite mass matrix, we employ the Crout algorithm. The performance and convergence of our method are rigorously evaluated through several benchmark numerical tests. Extensive comparisons with exact solutions and alternative methods demonstrate that our approach delivers an accurate, stable, and efficient computational scheme for the Burgers’ equation.
- Open Access
- 16 August 2024
- Published by MDPI AG in Mathematics
- Vol. 12 (16) , 2533
Continuum–Discontinuum Bonded-Block Model for Simulating Mixed-Mode Fractures
In this study, an improved discontinuous deformation analysis method with sub-block strategy is introduced to numerically simulate mixed-mode fractures. This approach partitions the material domain into continuum and potential discontinuum regions, applying specialized modeling techniques to each. In the continuum region, penalty-like bonding springs are employed to glue the sub-blocks together to capture the elastic behavior of the material. In the potential discontinuum region, the cohesive springs with the stiffness based on the cohesive zone model are implemented between sub-blocks to reproduce the process of crack nucleation and propagation. The primary advantage of this method is its capability to effectively model the transition of quasi-brittle solids from a continuous to a discontinuous stage through the degradation of cohesive springs. This accurately represents material failure while maintaining stability and consistency along uncracked interfaces. Another significant benefit is the method’s efficiency, as it avoids complex contact operations along sub-block interfaces before the cohesive spring between them fails. Validation through various benchmark numerical examples, such as cantilever beam-bending and diverse fracture simulations, demonstrates the method’s accuracy and robustness by comparing the results with analytical solutions. These comparisons show that the proposed method effectively captures the interplay between tensile and shear traction components in the mixed-mode crack propagation process.
- 16 August 2024
- Published by MDPI AG in Mathematics
- Vol. 12 (16) , 2533
Open AccessContinuum–Discontinuum Bonded-Block Model for Simulating Mixed-Mode Fractures
In this study, an improved discontinuous deformation analysis method with sub-block strategy is introduced to numerically simulate mixed-mode fractures. This approach partitions the material domain into continuum and potential discontinuum regions, applying specialized modeling techniques to each. In the continuum region, penalty-like bonding springs are employed to glue the sub-blocks together to capture the elastic behavior of the material. In the potential discontinuum region, the cohesive springs with the stiffness based on the cohesive zone model are implemented between sub-blocks to reproduce the process of crack nucleation and propagation. The primary advantage of this method is its capability to effectively model the transition of quasi-brittle solids from a continuous to a discontinuous stage through the degradation of cohesive springs. This accurately represents material failure while maintaining stability and consistency along uncracked interfaces. Another significant benefit is the method’s efficiency, as it avoids complex contact operations along sub-block interfaces before the cohesive spring between them fails. Validation through various benchmark numerical examples, such as cantilever beam-bending and diverse fracture simulations, demonstrates the method’s accuracy and robustness by comparing the results with analytical solutions. These comparisons show that the proposed method effectively captures the interplay between tensile and shear traction components in the mixed-mode crack propagation process.