Chengyi Liu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Chengyi Liu has a total of 31 co-authors. Here are the top co-authors by number of shared publications:
- Xianyong Gong
Institute of Geospatial Information, Information Engineering University, Zhengzhou 450001, China
9 shared publications - Fang Wu
Institute of Geospatial Information, Information Engineering University, Zhengzhou, China
7 shared publications - Yue Qiu
Institute of Geospatial Information, Information Engineering University, Zhengzhou, China
5 shared publications - Jichong Yin
Institute of Geospatial Information, Information Engineering University, Zhengzhou 450001, China
4 shared publications - Andong Wang
Institute of Surveying and Mapping, Information Engineering University, Avenue of Sciences, Zhengzhou 450001, China
3 shared publications Haizhong Qian
3 shared publications
Latest Publications
- Open Access
- 17 September 2026
- Published by Taylor & Francis in Geocarto International
- Vol. 41 (1)
Gated Sparse Graph Transformer integrating topographic and lithological features for drainage pattern recognition
1. Drainage networks are fundamental map components that influence the representation of other geographic features during map generalisation (Shu et al. 2013; Liu et al. 2015). Shaped by tectonic a...
- 17 September 2026
- Published by Taylor & Francis in Geocarto International
- Vol. 41 (1)
Open AccessGated Sparse Graph Transformer integrating topographic and lithological features for drainage pattern recognition
1. Drainage networks are fundamental map components that influence the representation of other geographic features during map generalisation (Shu et al. 2013; Liu et al. 2015). Shaped by tectonic a...
- Open Access
- 9 August 2026
- Published by MDPI AG in ISPRS International Journal of Geo-Information
- Vol. 15 (8) , 358
An Intelligent Incremental Update Method for Building Data Across Multiple Scales Supported by Categorical Boosting
Leveraging larger-scale data with higher currency to incrementally update smaller-scale data, thereby upholding consistency across multiple scale databases, has become a core focus of contemporary map production tasks centered on data updates. Existing methods rely on rule-based constraints to extract change information and identify update-required objects, which have notable limitations in terms of method generalization and constraints on results. To address the above issues, we propose an intelligent incremental updating method for different scale building datasets supported by Categorical Boosting (CatBoost). The proposed method forms a general incremental updating framework for buildings through three steps: change information extraction, change information classification, and change information updating. Experiments conducted on different scale datasets from Ningbo, China, demonstrate that the proposed method can effectively identify update-required objects and generate more reasonable updated smaller-scale data than the comparative methods.
- 9 August 2026
- Published by MDPI AG in ISPRS International Journal of Geo-Information
- Vol. 15 (8) , 358
Open AccessAn Intelligent Incremental Update Method for Building Data Across Multiple Scales Supported by Categorical Boosting
Leveraging larger-scale data with higher currency to incrementally update smaller-scale data, thereby upholding consistency across multiple scale databases, has become a core focus of contemporary map production tasks centered on data updates. Existing methods rely on rule-based constraints to extract change information and identify update-required objects, which have notable limitations in terms of method generalization and constraints on results. To address the above issues, we propose an intelligent incremental updating method for different scale building datasets supported by Categorical Boosting (CatBoost). The proposed method forms a general incremental updating framework for buildings through three steps: change information extraction, change information classification, and change information updating. Experiments conducted on different scale datasets from Ningbo, China, demonstrate that the proposed method can effectively identify update-required objects and generate more reasonable updated smaller-scale data than the comparative methods.
- Open Access
- 5 August 2026
- Published by Elsevier in International Journal of Applied Earth Observation and Geoinformation
- Vol. 153
V-BUILD: A graph Transformer-based decision framework for selective updating of vector building data
- 5 August 2026
- Published by Elsevier in International Journal of Applied Earth Observation and Geoinformation
- Vol. 153
Open AccessV-BUILD: A graph Transformer-based decision framework for selective updating of vector building data
- Open Access
- 1 January 2026
- Published by Walter de Gruyter GmbH in Open Geosciences
- Vol. 18 (1)
A method of island linear pattern recognition under the support of gestalt principle
Island linear pattern is an important spatial distribution pattern reflecting the natural geographic characteristics, and the identification of island linear pattern helps to mine the implicit spatial knowledge and enhance the existing data. Aiming at the distribution characteristics of the island linear pattern, which is overall continuous and locally fragmented, an island linear recognition method supported by the Gestalt principle is proposed. First, in order to reduce the adverse effect of broken distribution of islands on island linear pattern recognition, the aggregation degree clustering algorithm is used to identify locally aggregated island clusters, and the convex hull algorithm is used to permute the clusters as a whole. Then, according to the proximity, directionality, and wholeness principles of the Gestalt principle, the corresponding characteristic parameters and filtering rules are designed to achieve the identification of the island linear pattern. Finally, we construct the proximity graph with islands and clusters as units by Minimum Spanning Tree method (MST) to identify island linear pattern, and merge these results according to the principle of “large-scale priority”. The experiment proves that this method can overcome the negative influence of the dispersed distribution of islands on the pattern recognition, and effectively identify the island linear pattern from the overall and local levels. Furthermore, this method also plays a certain role in building linear pattern recognition.
- 1 January 2026
- Published by Walter de Gruyter GmbH in Open Geosciences
- Vol. 18 (1)
Open AccessA method of island linear pattern recognition under the support of gestalt principle
Island linear pattern is an important spatial distribution pattern reflecting the natural geographic characteristics, and the identification of island linear pattern helps to mine the implicit spatial knowledge and enhance the existing data. Aiming at the distribution characteristics of the island linear pattern, which is overall continuous and locally fragmented, an island linear recognition method supported by the Gestalt principle is proposed. First, in order to reduce the adverse effect of broken distribution of islands on island linear pattern recognition, the aggregation degree clustering algorithm is used to identify locally aggregated island clusters, and the convex hull algorithm is used to permute the clusters as a whole. Then, according to the proximity, directionality, and wholeness principles of the Gestalt principle, the corresponding characteristic parameters and filtering rules are designed to achieve the identification of the island linear pattern. Finally, we construct the proximity graph with islands and clusters as units by Minimum Spanning Tree method (MST) to identify island linear pattern, and merge these results according to the principle of “large-scale priority”. The experiment proves that this method can overcome the negative influence of the dispersed distribution of islands on the pattern recognition, and effectively identify the island linear pattern from the overall and local levels. Furthermore, this method also plays a certain role in building linear pattern recognition.
- Open Access
- 17 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 63 (01962892) , 1-22
TriadMapper: A Necessity–Diversity–Sensitivity Aware Network for Building Vector Mapping
Building vectorization from remote sensing imagery is a critical task challenged by three persistent issues: distinguishing buildings from complex surroundings, adapting to vast scale variations, and delineating intricate boundaries. While existing methods have addressed these issues with specific techniques, they often lack a unified framework that systematically integrates these capabilities. Inspired by the hierarchical process of human cartography, which progresses from a coarse overview to fine-grained refinement, we propose a guiding cognitive model for network design. This model decomposes the task into three sequential principles: Necessity, establishing foundational scene understanding with a sufficient receptive field; Diversity, adapting feature representation to targets of varying sizes; and Sensitivity, focusing on critical boundary details. Based on this model, we introduce TriadMapper, a network for high-precision building vectorization. TriadMapper implements this cognitive hierarchy through three integrated components: (i) it strategically adopts a large-kernel mechanism, which we term the Context-Aware Selective Kernel (CA-SK), to establish the necessary broad contextual awareness; (ii) it introduces a novel Global-Local Perception (GLP) module that adaptively fuses multi-scale features to address the diversity of building scales and shapes; and (iii) it proposes a novel Scene Adaptive Optimization (SAO) module that uses deformable attention for sensitive boundary localization. Furthermore, our Mask-guided Vertex Selection and Optimization (MVSO) strategy employs adaptive geometric thresholds for high-quality polygon generation. Comprehensive experiments on the AICrowd benchmark demonstrate state-of-the-art performance, achieving 81.6% AP and 0.680 PoLiS. TriadMapper validates that a framework guided by cognitive principles can be a robust and precise solution for building vectorization. Code is available at: https://github.com/ParkourX/TriadMapper.
- 17 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 63 (01962892) , 1-22
Open AccessTriadMapper: A Necessity–Diversity–Sensitivity Aware Network for Building Vector Mapping
Building vectorization from remote sensing imagery is a critical task challenged by three persistent issues: distinguishing buildings from complex surroundings, adapting to vast scale variations, and delineating intricate boundaries. While existing methods have addressed these issues with specific techniques, they often lack a unified framework that systematically integrates these capabilities. Inspired by the hierarchical process of human cartography, which progresses from a coarse overview to fine-grained refinement, we propose a guiding cognitive model for network design. This model decomposes the task into three sequential principles: Necessity, establishing foundational scene understanding with a sufficient receptive field; Diversity, adapting feature representation to targets of varying sizes; and Sensitivity, focusing on critical boundary details. Based on this model, we introduce TriadMapper, a network for high-precision building vectorization. TriadMapper implements this cognitive hierarchy through three integrated components: (i) it strategically adopts a large-kernel mechanism, which we term the Context-Aware Selective Kernel (CA-SK), to establish the necessary broad contextual awareness; (ii) it introduces a novel Global-Local Perception (GLP) module that adaptively fuses multi-scale features to address the diversity of building scales and shapes; and (iii) it proposes a novel Scene Adaptive Optimization (SAO) module that uses deformable attention for sensitive boundary localization. Furthermore, our Mask-guided Vertex Selection and Optimization (MVSO) strategy employs adaptive geometric thresholds for high-quality polygon generation. Comprehensive experiments on the AICrowd benchmark demonstrate state-of-the-art performance, achieving 81.6% AP and 0.680 PoLiS. TriadMapper validates that a framework guided by cognitive principles can be a robust and precise solution for building vectorization. Code is available at: https://github.com/ParkourX/TriadMapper.
- Open Access
- 13 February 2025
- Published by MDPI AG in ISPRS International Journal of Geo-Information
- Vol. 14 (2) , 83
4SIM: A Novel Description Model for the Ternary Spatial Relation “Between” of Buildings
Natural language spatial relations are often ambiguous and polysemic. They are also pluralistic and subjective in nature. Their descriptive methods are crucial but difficult in cartography, geographic information science, and map generalization. “Between” is a context-concerned spatial concept which is widely used to describe the arrangement of spatial objects. It involves the spatial distribution of at least three spatial objects and describes a scenario in which one (or more) object(s) is surrounded by objects on both sides. Existing models based on RCC and the n-intersection model are mainly used to describe binary spatial logic and are inadequate in describing the “between” ternary relation effectively. At present, although existing models can describe and reason about such ternary spatial relations, their limitations still exist. The description capability of ternary spatial relations is unspecific and polysemic, and certain results are inconsistent with spatial cognition and perception and application requirements. Therefore, this paper proposes a 4 Sides Intersection Model (4SIM) to express the spatial relation among ternary buildings in detail. Theoretically, 4SIM can effectively describe 45 spatial relations among ternary objects through the topological distances among the four boundaries of the region formed by the intermediate object and the two adjacent objects. The 4SIM has been demonstrated to offer a superior degree of accuracy in the depiction of spatial relations in comparison to the extant RIM method, thus providing new possibilities for map generalization.
- 13 February 2025
- Published by MDPI AG in ISPRS International Journal of Geo-Information
- Vol. 14 (2) , 83
Open Access4SIM: A Novel Description Model for the Ternary Spatial Relation “Between” of Buildings
Natural language spatial relations are often ambiguous and polysemic. They are also pluralistic and subjective in nature. Their descriptive methods are crucial but difficult in cartography, geographic information science, and map generalization. “Between” is a context-concerned spatial concept which is widely used to describe the arrangement of spatial objects. It involves the spatial distribution of at least three spatial objects and describes a scenario in which one (or more) object(s) is surrounded by objects on both sides. Existing models based on RCC and the n-intersection model are mainly used to describe binary spatial logic and are inadequate in describing the “between” ternary relation effectively. At present, although existing models can describe and reason about such ternary spatial relations, their limitations still exist. The description capability of ternary spatial relations is unspecific and polysemic, and certain results are inconsistent with spatial cognition and perception and application requirements. Therefore, this paper proposes a 4 Sides Intersection Model (4SIM) to express the spatial relation among ternary buildings in detail. Theoretically, 4SIM can effectively describe 45 spatial relations among ternary objects through the topological distances among the four boundaries of the region formed by the intermediate object and the two adjacent objects. The 4SIM has been demonstrated to offer a superior degree of accuracy in the depiction of spatial relations in comparison to the extant RIM method, thus providing new possibilities for map generalization.
- Open Access
- 20 April 2023
- Published by Wiley in Transactions in GIS
- Vol. 27 (3) , 752-776
Identification of drainage patterns using a graph convolutional neural network
Various geological factors shape drainage patterns. Identifying drainage patterns is a classic problem in topographical knowledge mining and map generalization. Existing rule‐based methods rely heavily on the parameter settings of cartographers for drainage‐pattern recognition. These methods effectively identify drainage patterns in specific areas but require manual parameter tuning to identify drainage patterns in other areas. Owing to the complexity of topological and geometric characteristics, drainage pattern recognition involves nonlinear problems, and it is difficult to build mapping relationships between characteristics and patterns using rule‐based methods. Therefore, we proposed a data‐driven method based on a graph convolutional neural network to avoid heavy reliance on human experience and automatically mine implicit relationships between characteristics and drainage patterns. First, six typical drainage patterns (dendritic, rectangular, parallel, trellis, reticulate, and fanned) were listed based on map specifications, and the unique characteristics of each drainage pattern were illustrated. Subsequently, the drainage graphs were constructed. The characteristics of the whole, local, and individual units in the drainage networks were quantified based on drainage vector data. Finally, an identification model was developed using graph convolution, self‐attention pooling, and multiple fully connected layers for drainage pattern recognition. After training and testing, the accuracy of our model (0.801 ± 0.014) was better than that of the rule‐based method (0.572 ± 0.000) and the traditional machine learning methods (less than 0.733 ± 0.016). The results demonstrate that the ability of our model to identify drainage patterns surpasses that of other methods.
- 20 April 2023
- Published by Wiley in Transactions in GIS
- Vol. 27 (3) , 752-776
Open AccessIdentification of drainage patterns using a graph convolutional neural network
Various geological factors shape drainage patterns. Identifying drainage patterns is a classic problem in topographical knowledge mining and map generalization. Existing rule‐based methods rely heavily on the parameter settings of cartographers for drainage‐pattern recognition. These methods effectively identify drainage patterns in specific areas but require manual parameter tuning to identify drainage patterns in other areas. Owing to the complexity of topological and geometric characteristics, drainage pattern recognition involves nonlinear problems, and it is difficult to build mapping relationships between characteristics and patterns using rule‐based methods. Therefore, we proposed a data‐driven method based on a graph convolutional neural network to avoid heavy reliance on human experience and automatically mine implicit relationships between characteristics and drainage patterns. First, six typical drainage patterns (dendritic, rectangular, parallel, trellis, reticulate, and fanned) were listed based on map specifications, and the unique characteristics of each drainage pattern were illustrated. Subsequently, the drainage graphs were constructed. The characteristics of the whole, local, and individual units in the drainage networks were quantified based on drainage vector data. Finally, an identification model was developed using graph convolution, self‐attention pooling, and multiple fully connected layers for drainage pattern recognition. After training and testing, the accuracy of our model (0.801 ± 0.014) was better than that of the rule‐based method (0.572 ± 0.000) and the traditional machine learning methods (less than 0.733 ± 0.016). The results demonstrate that the ability of our model to identify drainage patterns surpasses that of other methods.
- Open Access
- 24 December 2022
- Published by MDPI AG in Remote Sensing
- Vol. 15 (1) , 95
AFL-Net: Attentional Feature Learning Network for Building Extraction from Remote Sensing Images
Convolutional neural networks (CNNs) perform well in tasks of segmenting buildings from remote sensing images. However, the intraclass heterogeneity of buildings is high in images, while the interclass homogeneity between buildings and other nonbuilding objects is low. This leads to an inaccurate distinction between buildings and complex backgrounds. To overcome this challenge, we propose an Attentional Feature Learning Network (AFL-Net) that can accurately extract buildings from remote sensing images. We designed an attentional multiscale feature fusion (AMFF) module and a shape feature refinement (SFR) module to improve building recognition accuracy in complex environments. The AMFF module adaptively adjusts the weights of multi-scale features through the attention mechanism, which enhances the global perception and ensures the integrity of building segmentation results. The SFR module captures the shape features of the buildings, which enhances the network capability for identifying the area between building edges and surrounding nonbuilding objects and reduces the over-segmentation of buildings. An ablation study was conducted with both qualitative and quantitative analyses, verifying the effectiveness of the AMFF and SFR modules. The proposed AFL-Net achieved 91.37, 82.10, 73.27, and 79.81% intersection over union (IoU) values on the WHU Building Aerial Imagery, Inria Aerial Image Labeling, Massachusetts Buildings, and Building Instances of Typical Cities in China datasets, respectively. Thus, the AFL-Net offers the prospect of application for successful extraction of buildings from remote sensing images.
- 24 December 2022
- Published by MDPI AG in Remote Sensing
- Vol. 15 (1) , 95
Open AccessAFL-Net: Attentional Feature Learning Network for Building Extraction from Remote Sensing Images
Convolutional neural networks (CNNs) perform well in tasks of segmenting buildings from remote sensing images. However, the intraclass heterogeneity of buildings is high in images, while the interclass homogeneity between buildings and other nonbuilding objects is low. This leads to an inaccurate distinction between buildings and complex backgrounds. To overcome this challenge, we propose an Attentional Feature Learning Network (AFL-Net) that can accurately extract buildings from remote sensing images. We designed an attentional multiscale feature fusion (AMFF) module and a shape feature refinement (SFR) module to improve building recognition accuracy in complex environments. The AMFF module adaptively adjusts the weights of multi-scale features through the attention mechanism, which enhances the global perception and ensures the integrity of building segmentation results. The SFR module captures the shape features of the buildings, which enhances the network capability for identifying the area between building edges and surrounding nonbuilding objects and reduces the over-segmentation of buildings. An ablation study was conducted with both qualitative and quantitative analyses, verifying the effectiveness of the AMFF and SFR modules. The proposed AFL-Net achieved 91.37, 82.10, 73.27, and 79.81% intersection over union (IoU) values on the WHU Building Aerial Imagery, Inria Aerial Image Labeling, Massachusetts Buildings, and Building Instances of Typical Cities in China datasets, respectively. Thus, the AFL-Net offers the prospect of application for successful extraction of buildings from remote sensing images.
- Open Access
- 22 September 2022
- Published by MDPI AG in Remote Sensing
- Vol. 14 (19) , 4744
A Multiscale and Multitask Deep Learning Framework for Automatic Building Extraction
Detecting buildings, segmenting building footprints, and extracting building edges from high-resolution remote sensing images are vital in applications such as urban planning, change detection, smart cities, and map-making and updating. The tasks of building detection, footprint segmentation, and edge extraction affect each other to a certain extent. However, most previous works have focused on one of these three tasks and have lacked a multitask learning framework that can simultaneously solve the tasks of building detection, footprint segmentation and edge extraction, making it difficult to obtain smooth and complete buildings. This study proposes a novel multiscale and multitask deep learning framework to consider the dependencies among building detection, footprint segmentation, and edge extraction while completing all three tasks. In addition, a multitask feature fusion module is introduced into the deep learning framework to increase the robustness of feature extraction. A multitask loss function is also introduced to balance the training losses among the various tasks to obtain the best training results. Finally, the proposed method is applied to open-source building datasets and large-scale high-resolution remote sensing images and compared with other advanced building extraction methods. To verify the effectiveness of multitask learning, the performance of multitask learning and single-task training is compared in ablation experiments. The experimental results show that the proposed method has certain advantages over other methods and that multitask learning can effectively improve single-task performance.
- 22 September 2022
- Published by MDPI AG in Remote Sensing
- Vol. 14 (19) , 4744
Open AccessA Multiscale and Multitask Deep Learning Framework for Automatic Building Extraction
Detecting buildings, segmenting building footprints, and extracting building edges from high-resolution remote sensing images are vital in applications such as urban planning, change detection, smart cities, and map-making and updating. The tasks of building detection, footprint segmentation, and edge extraction affect each other to a certain extent. However, most previous works have focused on one of these three tasks and have lacked a multitask learning framework that can simultaneously solve the tasks of building detection, footprint segmentation and edge extraction, making it difficult to obtain smooth and complete buildings. This study proposes a novel multiscale and multitask deep learning framework to consider the dependencies among building detection, footprint segmentation, and edge extraction while completing all three tasks. In addition, a multitask feature fusion module is introduced into the deep learning framework to increase the robustness of feature extraction. A multitask loss function is also introduced to balance the training losses among the various tasks to obtain the best training results. Finally, the proposed method is applied to open-source building datasets and large-scale high-resolution remote sensing images and compared with other advanced building extraction methods. To verify the effectiveness of multitask learning, the performance of multitask learning and single-task training is compared in ablation experiments. The experimental results show that the proposed method has certain advantages over other methods and that multitask learning can effectively improve single-task performance.
- Open Access
- 12 August 2022
- Published by MDPI AG in Remote Sensing
- Vol. 14 (16) , 3914
MSL-Net: An Efficient Network for Building Extraction from Aerial Imagery
There remains several challenges that are encountered in the task of extracting buildings from aerial imagery using convolutional neural networks (CNNs). First, the tremendous complexity of existing building extraction networks impedes their practical application. In addition, it is arduous for networks to sufficiently utilize the various building features in different images. To address these challenges, we propose an efficient network called MSL-Net that focuses on both multiscale building features and multilevel image features. First, we use depthwise separable convolution (DSC) to significantly reduce the network complexity, and then we embed a group normalization (GN) layer in the inverted residual structure to alleviate network performance degradation. Furthermore, we extract multiscale building features through an atrous spatial pyramid pooling (ASPP) module and apply long skip connections to establish long-distance dependence to fuse features at different levels of the given image. Finally, we add a deformable convolution network layer before the pixel classification step to enhance the feature extraction capability of MSL-Net for buildings with irregular shapes. The experimental results obtained on three publicly available datasets demonstrate that our proposed method achieves state-of-the-art accuracy with a faster inference speed than that of competing approaches. Specifically, the proposed MSL-Net achieves 90.4%, 81.1% and 70.9% intersection over union (IoU) values on the WHU Building Aerial Imagery dataset, Inria Aerial Image Labeling dataset and Massachusetts Buildings dataset, respectively, with an inference speed of 101.4 frames per second (FPS) for an input image of size 3 × 512 × 512 on an NVIDIA RTX 3090 GPU. With an excellent tradeoff between accuracy and speed, our proposed MSL-Net may hold great promise for use in building extraction tasks.
- 12 August 2022
- Published by MDPI AG in Remote Sensing
- Vol. 14 (16) , 3914
Open AccessMSL-Net: An Efficient Network for Building Extraction from Aerial Imagery
There remains several challenges that are encountered in the task of extracting buildings from aerial imagery using convolutional neural networks (CNNs). First, the tremendous complexity of existing building extraction networks impedes their practical application. In addition, it is arduous for networks to sufficiently utilize the various building features in different images. To address these challenges, we propose an efficient network called MSL-Net that focuses on both multiscale building features and multilevel image features. First, we use depthwise separable convolution (DSC) to significantly reduce the network complexity, and then we embed a group normalization (GN) layer in the inverted residual structure to alleviate network performance degradation. Furthermore, we extract multiscale building features through an atrous spatial pyramid pooling (ASPP) module and apply long skip connections to establish long-distance dependence to fuse features at different levels of the given image. Finally, we add a deformable convolution network layer before the pixel classification step to enhance the feature extraction capability of MSL-Net for buildings with irregular shapes. The experimental results obtained on three publicly available datasets demonstrate that our proposed method achieves state-of-the-art accuracy with a faster inference speed than that of competing approaches. Specifically, the proposed MSL-Net achieves 90.4%, 81.1% and 70.9% intersection over union (IoU) values on the WHU Building Aerial Imagery dataset, Inria Aerial Image Labeling dataset and Massachusetts Buildings dataset, respectively, with an inference speed of 101.4 frames per second (FPS) for an input image of size 3 × 512 × 512 on an NVIDIA RTX 3090 GPU. With an excellent tradeoff between accuracy and speed, our proposed MSL-Net may hold great promise for use in building extraction tasks.