Fang Lin
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Fang Lin has a total of 32 co-authors. Here are the top co-authors by number of shared publications:
- Chuanhong Jin
Zhejiang University
2 shared publications - Baian Ning
College of Artificial Intelligence & Low-Altitude Technology, South China Agricultural University, Guangzhou 510642, China
1 shared publication C.H. Jin
State Key Laboratory of Silicon Materials, School of Materials Science and Engineering, Zhejiang University, Hangzhou 310027, China
1 shared publication- Guanjin Xu
College of Electronic Engineering, South China Agricultural University, Guangzhou 510642, China
1 shared publication - Haitao Xu
College of Electronic Engineering/College of Artificial Intelligence, South China Agricultural University, Guangzhou 510642, China
1 shared publication - Haoxin Yu
College of Artificial Intelligence & Low-Altitude Technology, South China Agricultural University, Guangzhou 510642, China
1 shared publication
Latest Publications
- Open Access
- 29 April 2026
- Published by OAE Publishing Inc. in Microstructures
- Vol. 6 (3)
Wave-Reconstruction Generative Adversarial Networks (WRGAN): wave function predicted from a high-resolution electron microscopy images
High-resolution transmission electron microscopy (HRTEM) is indispensable for atomic-scale characterization yet fundamentally limited by the inherent phase loss in conventional detectors including CCD. To overcome this barrier, we propose Wave-Reconstruction Generative Adversarial Networks (WRGAN) that directly predict wave function amplitude and phase from single HRTEM images. Our physics-guided framework employs a Unet++ generator within a Generative Adversarial Networks (GAN) architecture via defining a physics-guided consistency loss. A key advantage is that WRGAN, trained solely on simulated data, demonstrates robust performance when directly applied to experimental images. Validation on experimental Nb8W9O47 image shows predicted amplitudes and phases closely match the groundtruth wave functions. Significantly, WRGAN successfully resolves upper and lower surface projections in noisy single-wall carbon nanotube (SWCNT) images, enabling near-atomic-resolution 3D reconstruction.
- 29 April 2026
- Published by OAE Publishing Inc. in Microstructures
- Vol. 6 (3)
Open AccessWave-Reconstruction Generative Adversarial Networks (WRGAN): wave function predicted from a high-resolution electron microscopy images
High-resolution transmission electron microscopy (HRTEM) is indispensable for atomic-scale characterization yet fundamentally limited by the inherent phase loss in conventional detectors including CCD. To overcome this barrier, we propose Wave-Reconstruction Generative Adversarial Networks (WRGAN) that directly predict wave function amplitude and phase from single HRTEM images. Our physics-guided framework employs a Unet++ generator within a Generative Adversarial Networks (GAN) architecture via defining a physics-guided consistency loss. A key advantage is that WRGAN, trained solely on simulated data, demonstrates robust performance when directly applied to experimental images. Validation on experimental Nb8W9O47 image shows predicted amplitudes and phases closely match the groundtruth wave functions. Significantly, WRGAN successfully resolves upper and lower surface projections in noisy single-wall carbon nanotube (SWCNT) images, enabling near-atomic-resolution 3D reconstruction.
- Open Access
An Unsupervised Image Stitching Framework via Joint Iterative Optimization of Deformation Estimation, Feature Registration, and Seamless Blending
Image stitching is a computational technique designed to align and seamlessly fuse multiple overlapping images into a single panoramic image with an extended field of view. It plays a critical role in diverse domains, including mobile photography, autonomous navigation, and visual perception systems. However, most conventional image stitching pipelines implicitly assume that the input images have been pre-corrected for geometric distortions, particularly radial distortion inherent to wide-angle and fisheye lenses. This assumption often fails in practice, as many consumer-grade cameras lack built-in correction or calibration support. Consequently, applying standard image stitching methods to the uncorrected imagery frequently degrades feature correspondence reliability and introduces visible geometric misalignments and seam discontinuities in the final panorama. To overcome these limitations, this paper introduces a task-driven joint iterative optimization framework for image stitching that unifies unsupervised radial distortion correction, distortion-aware feature registration, and seam-aware blending within a single cohesive optimization objective. Specifically, lens distortion parameters are explicitly modeled as learnable variables and embedded into both the geometric registration and seam optimization sub-problems. An efficient closed-loop optimization strategy is then employed to jointly refine distortion parameters, homography estimates, and optimal seam paths in an alternating, mutually reinforcing manner. Implementation-wise, we first propose a calibration-free initial radial distortion estimation method which leverages intrinsic image gradients and epipolar consistency to provide physically plausible initialization for subsequent optimization. During iteration, distortion parameters are progressively refined by integrating robust geometric constraints derived from current feature matches (via RANSAC-based consensus filtering) with photometric consistency cues. Extensive experiments on multiple public benchmarks featuring pronounced radial distortion demonstrate that our method achieves superior stitching fidelity using metrics including PSNR and SSIM. It also confirms enhanced feature matching stability, which outperforms both distortion-agnostic approaches and two-stage pipelines that decouple distortion correction from registration. Furthermore, comprehensive ablation studies quantitatively and qualitatively validate the functional necessity and synergistic contribution of each core module, confirming the design rationale and effectiveness of the proposed joint optimization architecture.
Open AccessAn Unsupervised Image Stitching Framework via Joint Iterative Optimization of Deformation Estimation, Feature Registration, and Seamless Blending
Image stitching is a computational technique designed to align and seamlessly fuse multiple overlapping images into a single panoramic image with an extended field of view. It plays a critical role in diverse domains, including mobile photography, autonomous navigation, and visual perception systems. However, most conventional image stitching pipelines implicitly assume that the input images have been pre-corrected for geometric distortions, particularly radial distortion inherent to wide-angle and fisheye lenses. This assumption often fails in practice, as many consumer-grade cameras lack built-in correction or calibration support. Consequently, applying standard image stitching methods to the uncorrected imagery frequently degrades feature correspondence reliability and introduces visible geometric misalignments and seam discontinuities in the final panorama. To overcome these limitations, this paper introduces a task-driven joint iterative optimization framework for image stitching that unifies unsupervised radial distortion correction, distortion-aware feature registration, and seam-aware blending within a single cohesive optimization objective. Specifically, lens distortion parameters are explicitly modeled as learnable variables and embedded into both the geometric registration and seam optimization sub-problems. An efficient closed-loop optimization strategy is then employed to jointly refine distortion parameters, homography estimates, and optimal seam paths in an alternating, mutually reinforcing manner. Implementation-wise, we first propose a calibration-free initial radial distortion estimation method which leverages intrinsic image gradients and epipolar consistency to provide physically plausible initialization for subsequent optimization. During iteration, distortion parameters are progressively refined by integrating robust geometric constraints derived from current feature matches (via RANSAC-based consensus filtering) with photometric consistency cues. Extensive experiments on multiple public benchmarks featuring pronounced radial distortion demonstrate that our method achieves superior stitching fidelity using metrics including PSNR and SSIM. It also confirms enhanced feature matching stability, which outperforms both distortion-agnostic approaches and two-stage pipelines that decouple distortion correction from registration. Furthermore, comprehensive ablation studies quantitatively and qualitatively validate the functional necessity and synergistic contribution of each core module, confirming the design rationale and effectiveness of the proposed joint optimization architecture.
- Open Access
- 14 February 2023
- Published by Springer Nature in Scientific Reports
- Vol. 13 (1) , 1-9
A small-dataset-trained deep learning framework for identifying atoms on transmission electron microscopy images
To accurately identify atoms on noisy transmission electron microscope images, a deep learning (DL) approach is employed to estimate the map of probabilities at each pixel for being an atom with element discernment. Thanks to a delicately-designed loss function and the ability to extract features, the proposed DL networks can be trained by a small dataset created from approximately 30 experimental images, each with a size of 256 × 256 pixels2. The accuracy and robustness of the network were verified by resolving the structural defects of graphene and polar structures in PbTiO3/SrTiO3 multilayers from both the general TEM images and their imitated images on which intensities of some pixels lost randomly. Such a network has the potential to identify atoms from very few images of beam-sensitive material and explosive images recorded in a dynamical atomic process. The idea of using a small-dataset-trained DL framework to resolve a specific problem may prove instructive for practical DL applications in various fields.
- 14 February 2023
- Published by Springer Nature in Scientific Reports
- Vol. 13 (1) , 1-9
Open AccessA small-dataset-trained deep learning framework for identifying atoms on transmission electron microscopy images
To accurately identify atoms on noisy transmission electron microscope images, a deep learning (DL) approach is employed to estimate the map of probabilities at each pixel for being an atom with element discernment. Thanks to a delicately-designed loss function and the ability to extract features, the proposed DL networks can be trained by a small dataset created from approximately 30 experimental images, each with a size of 256 × 256 pixels2. The accuracy and robustness of the network were verified by resolving the structural defects of graphene and polar structures in PbTiO3/SrTiO3 multilayers from both the general TEM images and their imitated images on which intensities of some pixels lost randomly. Such a network has the potential to identify atoms from very few images of beam-sensitive material and explosive images recorded in a dynamical atomic process. The idea of using a small-dataset-trained DL framework to resolve a specific problem may prove instructive for practical DL applications in various fields.
- 1 July 2019
- Published by Elsevier in Ultramicroscopy
- Vol. 202, 114-120
CalAtom: A software for quantitatively analysing atomic columns in a transmission electron microscope image
- 1 July 2019
- Published by Elsevier in Ultramicroscopy
- Vol. 202, 114-120
CalAtom: A software for quantitatively analysing atomic columns in a transmission electron microscope image
- Open Access
Control of the Nucleation Density of Molybdenum Disulfide in Large-Scale Synthesis Using Chemical Vapor Deposition
Atmospheric pressure chemical vapor deposition (CVD) is presently a promising approach for preparing two-dimensional (2D) MoS2 crystals at high temperatures on SiO2/Si substrates. In this work, we propose an improved CVD method without hydrogen, which can increase formula flexibility by controlling the heating temperature of MoO3 powder and sulfur powder. The results show that the size and coverage of MoS2 domains vary largely, from discrete triangles to continuous film, on substrate. We find that the formation of MoS2 domains is dependent on the nucleation density of MoS2. Laminar flow theory is employed to elucidate the cause of the different shapes of MoS2 domains. The distribution of carrier gas speeds at the substrate surface leads to a change of nucleation density and a variation of domain morphology. Thus, nucleation density and domain morphology can be actively controlled by adjusting the carrier gas flow rate in the experimental system. These results are of significance for understanding the growth regulation of 2D MoS2 crystals.
Open AccessControl of the Nucleation Density of Molybdenum Disulfide in Large-Scale Synthesis Using Chemical Vapor Deposition
Atmospheric pressure chemical vapor deposition (CVD) is presently a promising approach for preparing two-dimensional (2D) MoS2 crystals at high temperatures on SiO2/Si substrates. In this work, we propose an improved CVD method without hydrogen, which can increase formula flexibility by controlling the heating temperature of MoO3 powder and sulfur powder. The results show that the size and coverage of MoS2 domains vary largely, from discrete triangles to continuous film, on substrate. We find that the formation of MoS2 domains is dependent on the nucleation density of MoS2. Laminar flow theory is employed to elucidate the cause of the different shapes of MoS2 domains. The distribution of carrier gas speeds at the substrate surface leads to a change of nucleation density and a variation of domain morphology. Thus, nucleation density and domain morphology can be actively controlled by adjusting the carrier gas flow rate in the experimental system. These results are of significance for understanding the growth regulation of 2D MoS2 crystals.
- 1 April 2011
- Published by Zhejiang University Press in Journal of Zhejiang University-SCIENCE A
- Vol. 12 (4) , 255-259
Physical properties of the junction of scandium and carbon nanotubes
Using first-principles calculations, the contact between the scandium (Sc) and semiconducting carbon nanotube (CNT) is investigated. This is one of the best quality of n-type contacts. Two junction models with (8,0) CNT on low-index Sc surfaces are constructed to elucidate the structural and electronic properties of Sc/CNT junctions. Analyses based on density of states and charge difference reveal that strong chemical bonds are formed between Sc and C atoms due to hybrid states of Sc 3d state and C 2π state. With respect to Ti(0001)/CNT junction, we find the dipole layer formed at the interface of Sc(0001)/CNT is comparable with that of Ti(0001)/CNT but gives a negative barrier at the interface. This indicates that the excellent contact properties of Sc metal electrode are caused by its low work function and excellent binding with CNT.
- 1 April 2011
- Published by Zhejiang University Press in Journal of Zhejiang University-SCIENCE A
- Vol. 12 (4) , 255-259
Physical properties of the junction of scandium and carbon nanotubes
Using first-principles calculations, the contact between the scandium (Sc) and semiconducting carbon nanotube (CNT) is investigated. This is one of the best quality of n-type contacts. Two junction models with (8,0) CNT on low-index Sc surfaces are constructed to elucidate the structural and electronic properties of Sc/CNT junctions. Analyses based on density of states and charge difference reveal that strong chemical bonds are formed between Sc and C atoms due to hybrid states of Sc 3d state and C 2π state. With respect to Ti(0001)/CNT junction, we find the dipole layer formed at the interface of Sc(0001)/CNT is comparable with that of Ti(0001)/CNT but gives a negative barrier at the interface. This indicates that the excellent contact properties of Sc metal electrode are caused by its low work function and excellent binding with CNT.
- 14 May 2009
- Published by American Physical Society (APS) in Physical Review Letters
- Vol. 102 (19) , 195505
Fabrication of a Freestanding Boron Nitride Single Layer and Its Defect Assignments
A freestanding single layer of hexagonal boron nitride (-BN) has been successfully fabricated by controlled energetic electron irradiation through a layer-by-layer sputtering process. We have successfully resolved atomic defects in -BN with triangle shapes by means of an aberration corrected high-resolution transmission electron microscopy with exit-wave reconstruction. Boron monovacancies are found to be preferably formed and the dominating zigzag-type edges are proved to be nitrogen terminated.- 14 May 2009
- Published by American Physical Society (APS) in Physical Review Letters
- Vol. 102 (19) , 195505
Fabrication of a Freestanding Boron Nitride Single Layer and Its Defect Assignments
A freestanding single layer of hexagonal boron nitride (-BN) has been successfully fabricated by controlled energetic electron irradiation through a layer-by-layer sputtering process. We have successfully resolved atomic defects in -BN with triangle shapes by means of an aberration corrected high-resolution transmission electron microscopy with exit-wave reconstruction. Boron monovacancies are found to be preferably formed and the dominating zigzag-type edges are proved to be nitrogen terminated.