Accurately identifying building distribution from remote sensing images with complex background information is challenging.The emergence of diffusion models has prompted the innovative idea of employing the reverse de...Accurately identifying building distribution from remote sensing images with complex background information is challenging.The emergence of diffusion models has prompted the innovative idea of employing the reverse denoising process to distill building distribution from these complex backgrounds.Building on this concept,we propose a novel framework,building extraction diffusion model(BEDiff),which meticulously refines the extraction of building footprints from remote sensing images in a stepwise fashion.Our approach begins with the design of booster guidance,a mechanism that extracts structural and semantic features from remote sensing images to serve as priors,thereby providing targeted guidance for the diffusion process.Additionally,we introduce a cross-feature fusion module(CFM)that bridges the semantic gap between different types of features,facilitating the integration of the attributes extracted by booster guidance into the diffusion process more effectively.Our proposed BEDiff marks the first application of diffusion models to the task of building extraction.Empirical evidence from extensive experiments on the Beijing building dataset demonstrates the superior performance of BEDiff,affirming its effectiveness and potential for enhancing the accuracy of building extraction in complex urban landscapes.展开更多
In the management of land resources and the protection of cultivated land,the law enforcement of land satellite images is often used as one of the main means.In recent years,the policies and regulations of the law enf...In the management of land resources and the protection of cultivated land,the law enforcement of land satellite images is often used as one of the main means.In recent years,the policies and regulations of the law enforcement of land satellite images have become more and more strict and been adjusted increasingly frequently,playing a decisive role in preventing excessive non-agricultural and non-food urbanization.In the process of the law enforcement,the extraction of suspected illegal buildings is the most important and time-consuming content.Compared with the traditional deep learning model,fully convolutional networks(FCN)has a great advantage in remote sensing image processing because its input images are not limited by size,and both convolution and deconvolution are independent of the overall size of images.In this paper,an intelligent extraction model of suspected illegal buildings from land satellite images based on deep learning FCN was built.Kaiyuan City,Yunnan Province was taken as an example.The verification results show that the global accuracy of this model was 86.6%in the process of building extraction,and mean intersection over union(mIoU)was 73.6%.This study can provide reference for the extraction of suspected illegal buildings in the law enforcement work of land satellite images,and reduce the tedious manual operation to a certain extent.展开更多
Building collapse is a significant cause of earthquake-related casualties; therefore, the rapid assessment of buildings damage is important for emergency management and rescue. Airborne light detection and ranging （L...Building collapse is a significant cause of earthquake-related casualties; therefore, the rapid assessment of buildings damage is important for emergency management and rescue. Airborne light detection and ranging （LiDAR） can acquire point cloud data in combination with height values, which in turn provides detailed information on building damage. However, the most previous approaches have used optical images and LiDAR data, or pre- and post-earthquake LiDAR data, to derive building damage information. This study applied surface normal algorithms to extract the degree of building damage. In this method, the angle between the surface normal and zenith （0） is used to identify damaged parts of a building, while the ratio of the standard deviation to the mean absolute deviation （σ/δ） of θ is used to obtain the degree of building damage. Quantitative analysis of 85 individual buildings with different roof types （i.e., flat top or pitched roofs） was conducted, and the results confirm that post-earthquake single LiDAR data are not affected by roof shape. Furthermore, the results confirm that θ is correlated to building damage, and that σ/δ represents an effective index to identify the degree of building damage.展开更多
Extraction of buildings from LIDAR data has been an active research field in recent years. A scheme for building detection and reconstruction from LIDAR data is presented with an object-oriented method which is based ...Extraction of buildings from LIDAR data has been an active research field in recent years. A scheme for building detection and reconstruction from LIDAR data is presented with an object-oriented method which is based on the buildings’ semantic rules. Two key steps are discussed: how to group the discrete LIDAR points into single objects and how to establish the buildings’ semantic rules. In the end, the buildings are reconstructed in 3D form and three common parametric building models (flat, gabled, hipped) are implemented.展开更多
As in many parts of the world, long-term excessive extraction of groundwater has caused significant land-surface sub- sidence in the residential areas of Datun coal mining district in East China. The recorded maximum ...As in many parts of the world, long-term excessive extraction of groundwater has caused significant land-surface sub- sidence in the residential areas of Datun coal mining district in East China. The recorded maximum level of subsidence in the area since 1976 to 2006 is 863 mm, and the area with an accumulative subsidence more than 200 mm has reached 33.1 km2 by the end of 2006. Over ten cases of building crack due to ground subsidence have already been observed. Spatial variation in ground subsi- dence often leads to a corresponding pattern of ground deformation. Buildings and underground infrastructures have been under a higher risk of damage in locations with greater differential ground deformation. Governmental guideline in China classifies build- ing damages into four different levels, based on the observable measures such as the width of wall crack, the degree of door and window deformation, the degree of wall inclination and the degree of structural destruction. Building damage level (BDL) is esti- mated by means of ground deformation analysis in terms of variations in slope gradient and curvature. Ground deformation analysis in terms of variations in slope gradient has shown that the areas of BDL III and BDL II sites account for about 0.013 km2 and 0.284 km2 respectively in 2006, and the predicted areas of BDL (define this first) III and II sites will be about 0.029 km2 and 0.423 km2 respectively by 2010. The situation is getting worse as subsidence continues. That calls for effective strategies for subsidence miti- gation and damage reduction, in terms of sustainable groundwater extraction, enhanced monitoring and the establishment of early warning systems.展开更多
Segmentation and edge regulation are studied deeply to extract buildings fromDSM data produced in this paper. Building segmentation is the first step to extract buildings, anda new segmentation method-adaptive iterati...Segmentation and edge regulation are studied deeply to extract buildings fromDSM data produced in this paper. Building segmentation is the first step to extract buildings, anda new segmentation method-adaptive iterative segmentation considering rati-o mean square-is proposedto extract the contour of buildings effectively. A sub-image (such as 50X50 pixels) of the image isprocessed in sequence, the average gray level and its ratio mean square are calculated first, thenthreshold of the sub-image is selected by using iterative threshold segmentation. The current pixelis segmented according to the threshold, the average gray level and the ratio mean square of thesub-image. The edge points of the building are grouped according to the azimuth of neighbor points,and then the optimal azimuth of the points that belong to the same group can be calculated by usingline interpolation.展开更多
The paper presents a general paradigm of semiautomatic building extraction from aerial stereo image pair.In the semiautomatic extraction system,the building model is defined by selected roof type through human-machine...The paper presents a general paradigm of semiautomatic building extraction from aerial stereo image pair.In the semiautomatic extraction system,the building model is defined by selected roof type through human-machine interface and input the approximation of area where the extracted building exists.Then under the knowledge of the roof type,low-level and mid-level processing including edge detection,straight line segments extraction and line segments grouping are used to establish the initial geometrical model of the roof-top.However,the initial geometrical model is not so accurate in geometry.To attain accurate results,a general least squares adjustment integrating the linear templates matching model with geometrical constraints in object-space is applied to refine the initial geometrical model.The adjustment model integrating the straight edge pattern and 3D constraints together is a well-studied optimal and anti-noise method.After gaining proper initial values,this adjustment model can flexibly process extraction of kinds of roof types by changing or assembling the geometrical constraints in object-space.展开更多
Automatic extraction features and buildings in particular from digital images is one of the most complex and challenging task faced by computer vision and photogrammetric communities. Extracted buildings are required ...Automatic extraction features and buildings in particular from digital images is one of the most complex and challenging task faced by computer vision and photogrammetric communities. Extracted buildings are required for varieties of applications including urban planning, creation of GIS databases and development of urban city models for taxation. For decades, extraction of features has been done by photogrammetric methods using stereo plotters and digital work stations. Photogrammetric methods are tedious, manually operated and require well-trained personnel. In recent years, there has been emergence of high-resolution space borne images, which have disclosed a large number of new opportunities for medium and large-scale topographic mapping. In this paper, a semi-automatic method is introduced to extract buildings in planned and informal settlements in urban areas from high resolution imagery. The proposed method uses modified snakes model and radial casting algorithm to initialize snakes contours and refinement of building outlines. The extraction rate is 91 percent as demonstrated by examples over selected test areas. The potential, limitations and future work is discussed.展开更多
As one of the main geographical elements in urban areas,buildings are closely related to the development of the city.Therefore,how to quickly and accurately extract building information from remote sensing images is o...As one of the main geographical elements in urban areas,buildings are closely related to the development of the city.Therefore,how to quickly and accurately extract building information from remote sensing images is of great significance for urban map updating,urban planning and construction,etc.Extracting building information around power facilities,especially obtaining this information from high-resolution images,has become one of the current hot topics in remote sensing technology research.This study made full use of the characteristics of GF-2 satellite remote sensing images,adopted an object-oriented classification method,combined with multi-scale segmentation technology and CART classification algorithm,and successfully extracted the buildings in the study area.The research results showed that the overall classification accuracy reached 89.5%and the Kappa coefficient was 0.86.Using the object-oriented CART classification algorithm for building extraction could be closer to actual ground objects and had higher accuracy.The extraction of buildings in the city contributed to urban development planning and provided decision support for management.展开更多
Building outline extraction from segmented point clouds is a critical step of building footprint generation.Existing methods for this task are often based on the convex hull and α-shape algorithm.There are also some ...Building outline extraction from segmented point clouds is a critical step of building footprint generation.Existing methods for this task are often based on the convex hull and α-shape algorithm.There are also some methods using grids and Delaunay triangulation.The common challenge of these methods is the determination of proper parameters.While deep learning-based methods have shown promise in reducing the impact and dependence on parameter selection,their reliance on datasets with ground truth information limits the generalization of these methods.In this study,a novel unsupervised approach,called PH-shape,is proposed to address the aforementioned challenge.The methods of Persistence Homology(PH)and Fourier descriptor are introduced into the task of building outline extraction.The PH from the theory of topological data analysis supports the automatic and adaptive determination of proper buffer radius,thus enabling the parameter-adaptive extraction of building outlines through buffering and“inverse”buffering.The quantitative and qualitative experiment results on two datasets with different point densities demonstrate the effectiveness of the proposed approach in the face of various building types,interior boundaries,and the density variation in the point cloud data of one building.The PH-supported parameter adaptivity helps the proposed approach overcome the challenge of parameter determination and data variations and achieve reliable extraction of building outlines.展开更多
During the operation, maintenance and upkeep of concrete buildings, surface cracks are often regarded as important warning signs of potential damage. Their precise segmentation plays a key role in assessing the health...During the operation, maintenance and upkeep of concrete buildings, surface cracks are often regarded as important warning signs of potential damage. Their precise segmentation plays a key role in assessing the health of a building. Traditional manual inspection is subjective, inefficient and has safety hazards. In contrast, current mainstream computer vision–based crack segmentation methods still suffer from missed detections, false detections, and segmentation discontinuities. These problems are particularly evident when dealing with small cracks, complex backgrounds, and blurred boundaries. For this reason, this paper proposes a lightweight building surface crack segmentation method, HL-YOLO, based on YOLOv11n-seg, which integrates an attention mechanism and a dilation-wise residual structure. First, we design a lightweight backbone network, RCSAA-Net, which combines ResNet50, capable of multi-scale feature extraction, with a custom Channel-Spatial Aggregation Attention (CSAA) module. This design boosts the model’s capacity to extract features of fine cracks and complex backgrounds. Among them, the CSAA module enhances the model’s attention to critical crack areas by capturing global dependencies in feature maps. Secondly, we construct an enhanced Content-aware ReAssembly of FEatures (ProCARAFE) module. It introduces a larger receptive field and dynamic kernel generation mechanism to achieve the reconstruction and accurate restoration of crack edge details. Finally, a Dilation-wise Residual (DWR) structure is introduced to reconstruct the C3k2 modules in the neck. It enhances multi-scale feature extraction and long-range contextual information fusion capabilities through multi-rate depthwise dilated convolutions. The improved model’s superiority and generalization ability have been validated through experiments on the self-built dataset. Compared to the baseline model, HL-YOLO improves mean Average Precision at 0.5 IoU by 4.1%, and increases the mean Intersection over Union (mIoU) by 4.86%, with only 3.12 million parameters. These results indicate that HL-YOLO can efficiently and accurately identify cracks on building surfaces, meeting the demand for rapid detection and providing an effective technical solution for real-time crack monitoring.展开更多
Extracting building contours from aerial images is a fundamental task in remote sensing.Current building extraction methods cannot accurately extract building contour information and have errors in extracting small-sc...Extracting building contours from aerial images is a fundamental task in remote sensing.Current building extraction methods cannot accurately extract building contour information and have errors in extracting small-scale buildings.This paper introduces a novel dense feature iterative(DFI)fusion network,denoted as DFINet,for extracting building contours.The network uses a DFI decoder to fuse semantic information at different scales and learns the building contour knowledge,producing the last features through iterative fusion.The dense feature fusion(DFF)module combines features at multiple scales.We employ the contour reconstruction(CR)module to access the final predictions.Extensive experiments validate the effectiveness of the DFINet on two different remote sensing datasets,INRIA aerial image dataset and Wuhan University(WHU)building dataset.On the INRIA aerial image dataset,our method achieves the highest intersection over union(IoU),overall accuracy(OA)and F 1 scores compared to other state-of-the-art methods.展开更多
The features of the surface subsidence basin caused by mining in the deep-lying seams are associated with the width of the coal pillar between two working faces. If the pillar is wide, a waved subsidence basin will oc...The features of the surface subsidence basin caused by mining in the deep-lying seams are associated with the width of the coal pillar between two working faces. If the pillar is wide, a waved subsidence basin will occur on the surface. If the pillar is narrow, the maximum surface subsidence value will be very great. The influence of the interval pillar to the stress distribution caused by mining in the deep-lying seam was specially studied by using a three-dimensional finite-difference FLAC3D program. The techniques mining in the deep-lying seams under buildings were presented.展开更多
An advanced edge-based method of feature detection and extraction is developed for object description in digital images. It is useful for the comparison of different images of the same scene in aerial imagery, for des...An advanced edge-based method of feature detection and extraction is developed for object description in digital images. It is useful for the comparison of different images of the same scene in aerial imagery, for describing and recognizing categories, for automatic building extraction and for finding the mutual regions in image matching. The method includes directional filtering and searching for straight edge segments in every direction and scale, taking into account edge gradient signs. Line segments are ordered with respect to their orientation and average gradients in the region in question. These segments are used for the construction of an object descriptor. A hierarchical set of feature descriptors is developed, taking into consideration the proposed straight line segment detector. Comparative performance is evaluated on the noisy model and in real aerial and satellite imagery.展开更多
[目的/意义]科技文献复杂知识对象对科技文献中的深度知识内容进行细粒度、全面的知识表示,可有效支撑数智驱动的科学发现与知识发现,是重要的科技创新要素。[方法/过程]首先,通过轻量级本体构建、BRAT知识标注和Neo4j知识存储等步骤实...[目的/意义]科技文献复杂知识对象对科技文献中的深度知识内容进行细粒度、全面的知识表示,可有效支撑数智驱动的科学发现与知识发现,是重要的科技创新要素。[方法/过程]首先,通过轻量级本体构建、BRAT知识标注和Neo4j知识存储等步骤实现领域知识图谱构建,其次,本地化部署大语言模型ChatGLM2-6B并通过低秩适应(Low-Rank Adaptation,LoRA)技术微调模型,最后基于思维记忆(Memory of Thoughts,MOT)机制将知识图谱中的复杂知识注入提示中,通过与大语言模型的多轮问答从科技文献中抽取出复杂知识对象。[结果/结论]以有机太阳能电池(Organic Solar Cells,OSC)为例验证方法的有效性,结果表明融合知识图谱与大语言模型的抽取方法优于大语言模型单独支撑的抽取方法,在准确率P、召回率R和F1值3个指标上分别提升14.1%、10.3%和12.3%。知识图谱能够增强大语言模型对科技文献的复杂知识对象抽取能力,提升OSC领域的科技文献挖掘效率与准确性。展开更多
基金supported by the National Natural Science Foundation of China(Nos.61906168,62202429 and 62272267)the Zhejiang Provincial Natural Science Foundation of China(No.LY23F020023)the Construction of Hubei Provincial Key Laboratory for Intelligent Visual Monitoring of Hydropower Projects(No.2022SDSJ01)。
文摘Accurately identifying building distribution from remote sensing images with complex background information is challenging.The emergence of diffusion models has prompted the innovative idea of employing the reverse denoising process to distill building distribution from these complex backgrounds.Building on this concept,we propose a novel framework,building extraction diffusion model(BEDiff),which meticulously refines the extraction of building footprints from remote sensing images in a stepwise fashion.Our approach begins with the design of booster guidance,a mechanism that extracts structural and semantic features from remote sensing images to serve as priors,thereby providing targeted guidance for the diffusion process.Additionally,we introduce a cross-feature fusion module(CFM)that bridges the semantic gap between different types of features,facilitating the integration of the attributes extracted by booster guidance into the diffusion process more effectively.Our proposed BEDiff marks the first application of diffusion models to the task of building extraction.Empirical evidence from extensive experiments on the Beijing building dataset demonstrates the superior performance of BEDiff,affirming its effectiveness and potential for enhancing the accuracy of building extraction in complex urban landscapes.
文摘In the management of land resources and the protection of cultivated land,the law enforcement of land satellite images is often used as one of the main means.In recent years,the policies and regulations of the law enforcement of land satellite images have become more and more strict and been adjusted increasingly frequently,playing a decisive role in preventing excessive non-agricultural and non-food urbanization.In the process of the law enforcement,the extraction of suspected illegal buildings is the most important and time-consuming content.Compared with the traditional deep learning model,fully convolutional networks(FCN)has a great advantage in remote sensing image processing because its input images are not limited by size,and both convolution and deconvolution are independent of the overall size of images.In this paper,an intelligent extraction model of suspected illegal buildings from land satellite images based on deep learning FCN was built.Kaiyuan City,Yunnan Province was taken as an example.The verification results show that the global accuracy of this model was 86.6%in the process of building extraction,and mean intersection over union(mIoU)was 73.6%.This study can provide reference for the extraction of suspected illegal buildings in the law enforcement work of land satellite images,and reduce the tedious manual operation to a certain extent.
基金supported by the National Natural Science Foundation of China(Grant No.41404046)the World Bank GFDRR group for providing financial support to acquire the data
文摘Building collapse is a significant cause of earthquake-related casualties; therefore, the rapid assessment of buildings damage is important for emergency management and rescue. Airborne light detection and ranging （LiDAR） can acquire point cloud data in combination with height values, which in turn provides detailed information on building damage. However, the most previous approaches have used optical images and LiDAR data, or pre- and post-earthquake LiDAR data, to derive building damage information. This study applied surface normal algorithms to extract the degree of building damage. In this method, the angle between the surface normal and zenith （0） is used to identify damaged parts of a building, while the ratio of the standard deviation to the mean absolute deviation （σ/δ） of θ is used to obtain the degree of building damage. Quantitative analysis of 85 individual buildings with different roof types （i.e., flat top or pitched roofs） was conducted, and the results confirm that post-earthquake single LiDAR data are not affected by roof shape. Furthermore, the results confirm that θ is correlated to building damage, and that σ/δ represents an effective index to identify the degree of building damage.
基金Supported by the Key Laboratory of Geo Informatics of State Bureau of Surveying and Mapping.
文摘Extraction of buildings from LIDAR data has been an active research field in recent years. A scheme for building detection and reconstruction from LIDAR data is presented with an object-oriented method which is based on the buildings’ semantic rules. Two key steps are discussed: how to group the discrete LIDAR points into single objects and how to establish the buildings’ semantic rules. In the end, the buildings are reconstructed in 3D form and three common parametric building models (flat, gabled, hipped) are implemented.
文摘As in many parts of the world, long-term excessive extraction of groundwater has caused significant land-surface sub- sidence in the residential areas of Datun coal mining district in East China. The recorded maximum level of subsidence in the area since 1976 to 2006 is 863 mm, and the area with an accumulative subsidence more than 200 mm has reached 33.1 km2 by the end of 2006. Over ten cases of building crack due to ground subsidence have already been observed. Spatial variation in ground subsi- dence often leads to a corresponding pattern of ground deformation. Buildings and underground infrastructures have been under a higher risk of damage in locations with greater differential ground deformation. Governmental guideline in China classifies build- ing damages into four different levels, based on the observable measures such as the width of wall crack, the degree of door and window deformation, the degree of wall inclination and the degree of structural destruction. Building damage level (BDL) is esti- mated by means of ground deformation analysis in terms of variations in slope gradient and curvature. Ground deformation analysis in terms of variations in slope gradient has shown that the areas of BDL III and BDL II sites account for about 0.013 km2 and 0.284 km2 respectively in 2006, and the predicted areas of BDL (define this first) III and II sites will be about 0.029 km2 and 0.423 km2 respectively by 2010. The situation is getting worse as subsidence continues. That calls for effective strategies for subsidence miti- gation and damage reduction, in terms of sustainable groundwater extraction, enhanced monitoring and the establishment of early warning systems.
基金theNationalNaturalScienceFoundationofChina (No 40 2 0 1 0 35)
文摘Segmentation and edge regulation are studied deeply to extract buildings fromDSM data produced in this paper. Building segmentation is the first step to extract buildings, anda new segmentation method-adaptive iterative segmentation considering rati-o mean square-is proposedto extract the contour of buildings effectively. A sub-image (such as 50X50 pixels) of the image isprocessed in sequence, the average gray level and its ratio mean square are calculated first, thenthreshold of the sub-image is selected by using iterative threshold segmentation. The current pixelis segmented according to the threshold, the average gray level and the ratio mean square of thesub-image. The edge points of the building are grouped according to the azimuth of neighbor points,and then the optimal azimuth of the points that belong to the same group can be calculated by usingline interpolation.
文摘The paper presents a general paradigm of semiautomatic building extraction from aerial stereo image pair.In the semiautomatic extraction system,the building model is defined by selected roof type through human-machine interface and input the approximation of area where the extracted building exists.Then under the knowledge of the roof type,low-level and mid-level processing including edge detection,straight line segments extraction and line segments grouping are used to establish the initial geometrical model of the roof-top.However,the initial geometrical model is not so accurate in geometry.To attain accurate results,a general least squares adjustment integrating the linear templates matching model with geometrical constraints in object-space is applied to refine the initial geometrical model.The adjustment model integrating the straight edge pattern and 3D constraints together is a well-studied optimal and anti-noise method.After gaining proper initial values,this adjustment model can flexibly process extraction of kinds of roof types by changing or assembling the geometrical constraints in object-space.
文摘Automatic extraction features and buildings in particular from digital images is one of the most complex and challenging task faced by computer vision and photogrammetric communities. Extracted buildings are required for varieties of applications including urban planning, creation of GIS databases and development of urban city models for taxation. For decades, extraction of features has been done by photogrammetric methods using stereo plotters and digital work stations. Photogrammetric methods are tedious, manually operated and require well-trained personnel. In recent years, there has been emergence of high-resolution space borne images, which have disclosed a large number of new opportunities for medium and large-scale topographic mapping. In this paper, a semi-automatic method is introduced to extract buildings in planned and informal settlements in urban areas from high resolution imagery. The proposed method uses modified snakes model and radial casting algorithm to initialize snakes contours and refinement of building outlines. The extraction rate is 91 percent as demonstrated by examples over selected test areas. The potential, limitations and future work is discussed.
基金Research on Algorithm Model for Monitoring and Evaluating Typical Disaster Situations of Electric Power Equipment Based on Remote Sensing Imaging Technology of Heaven and Earth,South Grid Guangxi Power Grid Company Science and Technology Project(GXKJXM20222160).
文摘As one of the main geographical elements in urban areas,buildings are closely related to the development of the city.Therefore,how to quickly and accurately extract building information from remote sensing images is of great significance for urban map updating,urban planning and construction,etc.Extracting building information around power facilities,especially obtaining this information from high-resolution images,has become one of the current hot topics in remote sensing technology research.This study made full use of the characteristics of GF-2 satellite remote sensing images,adopted an object-oriented classification method,combined with multi-scale segmentation technology and CART classification algorithm,and successfully extracted the buildings in the study area.The research results showed that the overall classification accuracy reached 89.5%and the Kappa coefficient was 0.86.Using the object-oriented CART classification algorithm for building extraction could be closer to actual ground objects and had higher accuracy.The extraction of buildings in the city contributed to urban development planning and provided decision support for management.
基金supported by NTNU Digital project[grant number 81771593].
文摘Building outline extraction from segmented point clouds is a critical step of building footprint generation.Existing methods for this task are often based on the convex hull and α-shape algorithm.There are also some methods using grids and Delaunay triangulation.The common challenge of these methods is the determination of proper parameters.While deep learning-based methods have shown promise in reducing the impact and dependence on parameter selection,their reliance on datasets with ground truth information limits the generalization of these methods.In this study,a novel unsupervised approach,called PH-shape,is proposed to address the aforementioned challenge.The methods of Persistence Homology(PH)and Fourier descriptor are introduced into the task of building outline extraction.The PH from the theory of topological data analysis supports the automatic and adaptive determination of proper buffer radius,thus enabling the parameter-adaptive extraction of building outlines through buffering and“inverse”buffering.The quantitative and qualitative experiment results on two datasets with different point densities demonstrate the effectiveness of the proposed approach in the face of various building types,interior boundaries,and the density variation in the point cloud data of one building.The PH-supported parameter adaptivity helps the proposed approach overcome the challenge of parameter determination and data variations and achieve reliable extraction of building outlines.
基金support from Natural Science Foundation of Hunan Province(Grant No.2024JJ8055)Hunan Yiduoyun Commodity Itelligence Project(Grant No.h2024-003).
文摘During the operation, maintenance and upkeep of concrete buildings, surface cracks are often regarded as important warning signs of potential damage. Their precise segmentation plays a key role in assessing the health of a building. Traditional manual inspection is subjective, inefficient and has safety hazards. In contrast, current mainstream computer vision–based crack segmentation methods still suffer from missed detections, false detections, and segmentation discontinuities. These problems are particularly evident when dealing with small cracks, complex backgrounds, and blurred boundaries. For this reason, this paper proposes a lightweight building surface crack segmentation method, HL-YOLO, based on YOLOv11n-seg, which integrates an attention mechanism and a dilation-wise residual structure. First, we design a lightweight backbone network, RCSAA-Net, which combines ResNet50, capable of multi-scale feature extraction, with a custom Channel-Spatial Aggregation Attention (CSAA) module. This design boosts the model’s capacity to extract features of fine cracks and complex backgrounds. Among them, the CSAA module enhances the model’s attention to critical crack areas by capturing global dependencies in feature maps. Secondly, we construct an enhanced Content-aware ReAssembly of FEatures (ProCARAFE) module. It introduces a larger receptive field and dynamic kernel generation mechanism to achieve the reconstruction and accurate restoration of crack edge details. Finally, a Dilation-wise Residual (DWR) structure is introduced to reconstruct the C3k2 modules in the neck. It enhances multi-scale feature extraction and long-range contextual information fusion capabilities through multi-rate depthwise dilated convolutions. The improved model’s superiority and generalization ability have been validated through experiments on the self-built dataset. Compared to the baseline model, HL-YOLO improves mean Average Precision at 0.5 IoU by 4.1%, and increases the mean Intersection over Union (mIoU) by 4.86%, with only 3.12 million parameters. These results indicate that HL-YOLO can efficiently and accurately identify cracks on building surfaces, meeting the demand for rapid detection and providing an effective technical solution for real-time crack monitoring.
基金National Natural Science Foundation of China(No.61903078)Fundamental Research Funds for the Central Universities,China(No.2232021A-10)+1 种基金Shanghai Sailing Program,China(No.22YF1401300)Natural Science Foundation of Shanghai,China(No.20ZR1400400)。
文摘Extracting building contours from aerial images is a fundamental task in remote sensing.Current building extraction methods cannot accurately extract building contour information and have errors in extracting small-scale buildings.This paper introduces a novel dense feature iterative(DFI)fusion network,denoted as DFINet,for extracting building contours.The network uses a DFI decoder to fuse semantic information at different scales and learns the building contour knowledge,producing the last features through iterative fusion.The dense feature fusion(DFF)module combines features at multiple scales.We employ the contour reconstruction(CR)module to access the final predictions.Extensive experiments validate the effectiveness of the DFINet on two different remote sensing datasets,INRIA aerial image dataset and Wuhan University(WHU)building dataset.On the INRIA aerial image dataset,our method achieves the highest intersection over union(IoU),overall accuracy(OA)and F 1 scores compared to other state-of-the-art methods.
文摘The features of the surface subsidence basin caused by mining in the deep-lying seams are associated with the width of the coal pillar between two working faces. If the pillar is wide, a waved subsidence basin will occur on the surface. If the pillar is narrow, the maximum surface subsidence value will be very great. The influence of the interval pillar to the stress distribution caused by mining in the deep-lying seam was specially studied by using a three-dimensional finite-difference FLAC3D program. The techniques mining in the deep-lying seams under buildings were presented.
文摘An advanced edge-based method of feature detection and extraction is developed for object description in digital images. It is useful for the comparison of different images of the same scene in aerial imagery, for describing and recognizing categories, for automatic building extraction and for finding the mutual regions in image matching. The method includes directional filtering and searching for straight edge segments in every direction and scale, taking into account edge gradient signs. Line segments are ordered with respect to their orientation and average gradients in the region in question. These segments are used for the construction of an object descriptor. A hierarchical set of feature descriptors is developed, taking into consideration the proposed straight line segment detector. Comparative performance is evaluated on the noisy model and in real aerial and satellite imagery.
文摘[目的/意义]科技文献复杂知识对象对科技文献中的深度知识内容进行细粒度、全面的知识表示,可有效支撑数智驱动的科学发现与知识发现,是重要的科技创新要素。[方法/过程]首先,通过轻量级本体构建、BRAT知识标注和Neo4j知识存储等步骤实现领域知识图谱构建,其次,本地化部署大语言模型ChatGLM2-6B并通过低秩适应(Low-Rank Adaptation,LoRA)技术微调模型,最后基于思维记忆(Memory of Thoughts,MOT)机制将知识图谱中的复杂知识注入提示中,通过与大语言模型的多轮问答从科技文献中抽取出复杂知识对象。[结果/结论]以有机太阳能电池(Organic Solar Cells,OSC)为例验证方法的有效性,结果表明融合知识图谱与大语言模型的抽取方法优于大语言模型单独支撑的抽取方法,在准确率P、召回率R和F1值3个指标上分别提升14.1%、10.3%和12.3%。知识图谱能够增强大语言模型对科技文献的复杂知识对象抽取能力,提升OSC领域的科技文献挖掘效率与准确性。
