Hengjian Tong
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Hengjian Tong has a total of 17 co-authors. Here are the top co-authors by number of shared publications:
- Fei Tong
Department of Geodesy and Geomatics Engineering, University of New Brunswick, 15 Dineen Drive, Fredericton, NB E3B 5A3, Canada
4 shared publications - Yun Zhang
Department of Geodesy and Geomatics Engineering, University of New Brunswick, Fredericton, Canada, Canada
4 shared publications Bohua Li
School of Computer Science, China University of Geosciences, Wuhan, China
1 shared publication- Chang Tang
Huazhong University of Science and Technology
1 shared publication Hao Dou
Huazhong University of Science and Technology
1 shared publication- Jiechen Tang
School of Computer Science, China University of Geosciences, 68 Jincheng Street, East Lake New Technology Development Zone, Wuhan 430078, China
1 shared publication
Latest Publications
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Biology and Bioinformatics
- Vol. 23 (5) , 1-14
Boundary-Aware Clustering of Spatial Transcriptomics Data Via Fourier Feature Mapping and Dynamic Self-Supervision
Accurate identification of fine-grained spatial domains within tissues using spatial transcriptomics data is critical for understanding tissue development, disease pathology, and cellular interactions. However, inherent spatial heterogeneity in spatial transcriptomic data, characterized by significant discontinuity in spatial gradients of gene expression, makes it difficult for existing clustering methods to precisely delineate fine-grained structures and produce ambiguous spatial boundaries. To address this challenge, this paper proposes FPS-MGCN - a multi-view graph convolutional network integrating Fourier feature mapping with dynamic pseudo-label self-supervision. In detail, to tackle boundary ambiguity caused by spatial heterogeneity, we introduce a Fourier sine-cosine feature mapping that explicitly constructs high-frequency feature channels to enhance sensitivity to high-frequency spatial gradients such as tissue boundaries. In addition, a dynamic pseudo-label self-supervision strategy is further designed, providing progressive semantic constraints that smooth noise interference and enhance recognition in spatial boundary regions. The model integrates raw gene expression, Fourier-mapped features, and spatial adjacency information through a multi-view GCN encoder, employing an attention mechanism for adaptive fusion. Extensive experimental results demonstrate that the proposed FPS-MGCN identifies critical fine-grained structures difficult to resolve by traditional methods and outperforms other state-of-the-art methods in spatial domain identification.
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Biology and Bioinformatics
- Vol. 23 (5) , 1-14
Boundary-Aware Clustering of Spatial Transcriptomics Data Via Fourier Feature Mapping and Dynamic Self-Supervision
Accurate identification of fine-grained spatial domains within tissues using spatial transcriptomics data is critical for understanding tissue development, disease pathology, and cellular interactions. However, inherent spatial heterogeneity in spatial transcriptomic data, characterized by significant discontinuity in spatial gradients of gene expression, makes it difficult for existing clustering methods to precisely delineate fine-grained structures and produce ambiguous spatial boundaries. To address this challenge, this paper proposes FPS-MGCN - a multi-view graph convolutional network integrating Fourier feature mapping with dynamic pseudo-label self-supervision. In detail, to tackle boundary ambiguity caused by spatial heterogeneity, we introduce a Fourier sine-cosine feature mapping that explicitly constructs high-frequency feature channels to enhance sensitivity to high-frequency spatial gradients such as tissue boundaries. In addition, a dynamic pseudo-label self-supervision strategy is further designed, providing progressive semantic constraints that smooth noise interference and enhance recognition in spatial boundary regions. The model integrates raw gene expression, Fourier-mapped features, and spatial adjacency information through a multi-view GCN encoder, employing an attention mechanism for adaptive fusion. Extensive experimental results demonstrate that the proposed FPS-MGCN identifies critical fine-grained structures difficult to resolve by traditional methods and outperforms other state-of-the-art methods in spatial domain identification.
- Open Access
- 25 January 2023
- Published by MDPI AG in Remote Sensing
- Vol. 15 (3) , 715
Exploiting Superpixel-Based Contextual Information on Active Learning for High Spatial Resolution Remote Sensing Image Classification
Superpixel-based classification using Active Learning (AL) has shown great potential in high spatial resolution remote sensing image classification tasks. However, in existing superpixel-based classification models using AL, the expert labeling information is only used on the selected informative superpixel while its neighboring superpixels are ignored. Actually, as most superpixels are over-segmented, a ground object always contains multiple superpixels. Thus, the center superpixel tends to have the same label as its neighboring superpixels. In this paper, to make full use of the expert labeling information, a Similar Neighboring Superpixels Search and Labeling (SNSSL) method was proposed and used in the AL process. Firstly, we identify superpixels with certain categories and uncertain superpixels by supervised learning. Secondly, we use the active learning method to process those uncertain superpixels. In each round of AL, the expert labeling information is not only used to enrich the training set but also used to label the similar neighboring superpixels. Similar neighboring superpixels are determined by computing the similarity of two superpixels according to CIELAB Dominant Colors distance, Correlation distance, Angular Second Moment distance and Contrast distance. The final classification map is composed of the supervised learning classification map and the active learning with SNSSL classification map. To demonstrate the performance of the proposed SNSSL method, the experiments were conducted on images from two benchmark high spatial resolution remote sensing datasets. The experiment shows that overall accuracy, average accuracy and kappa coefficients of the classification using the SNSSL have been improved obviously compared with the classification without the SNSSL.
- 25 January 2023
- Published by MDPI AG in Remote Sensing
- Vol. 15 (3) , 715
Open AccessExploiting Superpixel-Based Contextual Information on Active Learning for High Spatial Resolution Remote Sensing Image Classification
Superpixel-based classification using Active Learning (AL) has shown great potential in high spatial resolution remote sensing image classification tasks. However, in existing superpixel-based classification models using AL, the expert labeling information is only used on the selected informative superpixel while its neighboring superpixels are ignored. Actually, as most superpixels are over-segmented, a ground object always contains multiple superpixels. Thus, the center superpixel tends to have the same label as its neighboring superpixels. In this paper, to make full use of the expert labeling information, a Similar Neighboring Superpixels Search and Labeling (SNSSL) method was proposed and used in the AL process. Firstly, we identify superpixels with certain categories and uncertain superpixels by supervised learning. Secondly, we use the active learning method to process those uncertain superpixels. In each round of AL, the expert labeling information is not only used to enrich the training set but also used to label the similar neighboring superpixels. Similar neighboring superpixels are determined by computing the similarity of two superpixels according to CIELAB Dominant Colors distance, Correlation distance, Angular Second Moment distance and Contrast distance. The final classification map is composed of the supervised learning classification map and the active learning with SNSSL classification map. To demonstrate the performance of the proposed SNSSL method, the experiments were conducted on images from two benchmark high spatial resolution remote sensing datasets. The experiment shows that overall accuracy, average accuracy and kappa coefficients of the classification using the SNSSL have been improved obviously compared with the classification without the SNSSL.
- Open Access
- 29 July 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 14 (19391404) , 7751-7761
Delineation of Individual Tree Crowns Using High Spatial Resolution Multispectral WorldView-3 Satellite Imagery
Delineation of individual tree crowns can provide valuable information for sustainable forest management and environmental protection. However, it is hard to find a reliable tree crown delineation method that can continually generate expected results in high spatial resolution multispectral satellite images, because most of the existing methods need user-assigned parameters that greatly affect the quality of the delineation results. In this article, we propose a method based on the marker-controlled watershed segmentation to delineate individual tree crowns using high spatial resolution multispectral WorldView-3 satellite imagery. A gradient binarization process is proposed to accurately locate tree crown borders. The threshold for the binarization is determined by a supervised searching process. Markers used in marker-controlled watershed segmentation are spatial local maxima detected from the information provided by tree crown borders. Moreover, the definition of spatial local maxima from the literature is improved to eliminate false treetops. To validate the performance of the proposed delineation method, delineation results are compared with those obtained from the spectral angle segmentation (SAS) method that has been proposed in literature because the quality of delineation results generated by SAS does not rely on user-assigned parameters. The experiment results in two test images demonstrate that the proposed method outperforms SAS in terms of both delineation accuracy and visual quality of the delineation map. Moreover, it is proved that the modified spatial local maxima are more reliable for detecting treetops.
- 29 July 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 14 (19391404) , 7751-7761
Open AccessDelineation of Individual Tree Crowns Using High Spatial Resolution Multispectral WorldView-3 Satellite Imagery
Delineation of individual tree crowns can provide valuable information for sustainable forest management and environmental protection. However, it is hard to find a reliable tree crown delineation method that can continually generate expected results in high spatial resolution multispectral satellite images, because most of the existing methods need user-assigned parameters that greatly affect the quality of the delineation results. In this article, we propose a method based on the marker-controlled watershed segmentation to delineate individual tree crowns using high spatial resolution multispectral WorldView-3 satellite imagery. A gradient binarization process is proposed to accurately locate tree crown borders. The threshold for the binarization is determined by a supervised searching process. Markers used in marker-controlled watershed segmentation are spatial local maxima detected from the information provided by tree crown borders. Moreover, the definition of spatial local maxima from the literature is improved to eliminate false treetops. To validate the performance of the proposed delineation method, delineation results are compared with those obtained from the spectral angle segmentation (SAS) method that has been proposed in literature because the quality of delineation results generated by SAS does not rely on user-assigned parameters. The experiment results in two test images demonstrate that the proposed method outperforms SAS in terms of both delineation accuracy and visual quality of the delineation map. Moreover, it is proved that the modified spatial local maxima are more reliable for detecting treetops.
- Open Access
- 10 November 2019
- Published by MDPI AG in Remote Sensing
- Vol. 11 (22) , 2627
Purifying SLIC Superpixels to Optimize Superpixel-Based Classification of High Spatial Resolution Remote Sensing Image
Fast and accurate classification of high spatial resolution remote sensing image is important for many applications. The usage of superpixels in classification has been proposed to accelerate the speed of classification. However, although most superpixels only contain pixels from single class, there are still some mixed superpixels, which mostly locate near the edge of different classes, and contain pixels from more than one class. Such mixed superpixels will cause misclassification regardless of classification methods used. In this paper, a superpixels purification algorithm based on color quantization is proposed to purify mixed Simple Linear Iterative Clustering (SLIC) superpixels. After purifying, the mixed SLIC superpixel will be separated into smaller superpixels. These smaller superpixels are pure superpixels which only contain a single kind of ground object. The experiments on images from the dataset BSDS500 show that the purified SLIC superpixels outperform the original SLIC superpixels on three segmentation evaluation metrics. With the purified SLIC superpixels, a classification scheme in which only edge superpixels are selected to be purified is proposed. The strategy of purifying edge superpixels not only improves the efficiency of the algorithm, but also improves the accuracy of the classification. The experiments on a remote sensing image from WorldView-2 satellite demonstrate that purified SLIC superpixels at all scales can generate classification result with higher accuracy than original SLIC superpixels, especially at the scale of , for which the accuracy increase is higher than 4%.- 10 November 2019
- Published by MDPI AG in Remote Sensing
- Vol. 11 (22) , 2627
Open AccessPurifying SLIC Superpixels to Optimize Superpixel-Based Classification of High Spatial Resolution Remote Sensing Image
Fast and accurate classification of high spatial resolution remote sensing image is important for many applications. The usage of superpixels in classification has been proposed to accelerate the speed of classification. However, although most superpixels only contain pixels from single class, there are still some mixed superpixels, which mostly locate near the edge of different classes, and contain pixels from more than one class. Such mixed superpixels will cause misclassification regardless of classification methods used. In this paper, a superpixels purification algorithm based on color quantization is proposed to purify mixed Simple Linear Iterative Clustering (SLIC) superpixels. After purifying, the mixed SLIC superpixel will be separated into smaller superpixels. These smaller superpixels are pure superpixels which only contain a single kind of ground object. The experiments on images from the dataset BSDS500 show that the purified SLIC superpixels outperform the original SLIC superpixels on three segmentation evaluation metrics. With the purified SLIC superpixels, a classification scheme in which only edge superpixels are selected to be purified is proposed. The strategy of purifying edge superpixels not only improves the efficiency of the algorithm, but also improves the accuracy of the classification. The experiments on a remote sensing image from WorldView-2 satellite demonstrate that purified SLIC superpixels at all scales can generate classification result with higher accuracy than original SLIC superpixels, especially at the scale of , for which the accuracy increase is higher than 4%. - Open Access
- 23 August 2017
- Published by MDPI AG in Remote Sensing
- Vol. 9 (9) , 872
Multiscale Union Regions Adaptive Sparse Representation for Hyperspectral Image Classification
Sparse Representation has been widely applied to classification of hyperspectral images (HSIs). Besides spectral information, the spatial context in HSIs also plays an important role in the classification. The recently published Multiscale Adaptive Sparse Representation (MASR) classifier has shown good performance in exploiting spatial information for HSI classification. But the spatial information is exploited by multiscale patches with fixed sizes of square windows. The patch can include all nearest neighbor pixels but these neighbor pixels may contain some noise pixels. Then another research proposed a Multiscale Superpixel-Based Sparse Representation (MSSR) classifier. Shape-adaptive superpixels can provide more accurate representation than patches. But it is difficult to select scales for superpixels. Therefore, inspired by the merits and demerits of multiscale patches and superpixels, we propose a novel algorithm called Multiscale Union Regions Adaptive Sparse Representation (MURASR). The union region, which is the overlap of patch and superpixel, can make full use of the advantages of both and overcome the weaknesses of each one. Experiments on several HSI datasets demonstrate that the proposed MURASR is superior to MASR and union region is better than the patch in the sparse representation.
- 23 August 2017
- Published by MDPI AG in Remote Sensing
- Vol. 9 (9) , 872
Open AccessMultiscale Union Regions Adaptive Sparse Representation for Hyperspectral Image Classification
Sparse Representation has been widely applied to classification of hyperspectral images (HSIs). Besides spectral information, the spatial context in HSIs also plays an important role in the classification. The recently published Multiscale Adaptive Sparse Representation (MASR) classifier has shown good performance in exploiting spatial information for HSI classification. But the spatial information is exploited by multiscale patches with fixed sizes of square windows. The patch can include all nearest neighbor pixels but these neighbor pixels may contain some noise pixels. Then another research proposed a Multiscale Superpixel-Based Sparse Representation (MSSR) classifier. Shape-adaptive superpixels can provide more accurate representation than patches. But it is difficult to select scales for superpixels. Therefore, inspired by the merits and demerits of multiscale patches and superpixels, we propose a novel algorithm called Multiscale Union Regions Adaptive Sparse Representation (MURASR). The union region, which is the overlap of patch and superpixel, can make full use of the advantages of both and overcome the weaknesses of each one. Experiments on several HSI datasets demonstrate that the proposed MURASR is superior to MASR and union region is better than the patch in the sparse representation.
- 7 December 2016
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 76 (21) , 22943-22958
Land-use classification with biologically inspired color descriptor and sparse coding spatial pyramid matching
- 7 December 2016
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 76 (21) , 22943-22958
Land-use classification with biologically inspired color descriptor and sparse coding spatial pyramid matching