AR
A. Rafiee
info
Please Note
<p>This page displays the records of the person named above and is not linked to a unique person identifier. This record may need to be merged to a profile.</p>
11 records found
1
Master thesis
(2026)
-
G. Iliopoulos, G.A.K. Arroyo Ohori, E. Brembilla, N. Forouzandeh Shahraki, A. Rafiee, H.W. de Wolff
This project evaluates the suitability of 3D interior space models acquired using Apple’s RoomPlan API for daylight simulations. The main contribution is a Python-based tool that converts RoomPlan output into HoneyBeeJSON by automatically reconstructing the ceiling and adding window frames, which are missing in RoomPlan’s output. Although RoomPlan is also able to capture furniture, these elements were not used in the geometric evaluation or in the daylight simulations. The resulting models can be directly used in Grasshopper for daylight simulations, reducing the modeling time required by practitioners. To assess the suitability of RoomPlan, three office interiors were scanned using a TLS and modeled both manually and with an iPhone 12 Pro.
The manual models were used as ground truth. For each room, a geometric evaluation and a daylight simulation evaluation were performed using three model versions: manual, RoomPlan with extruded window frames, and RoomPlan without extruded window frames. For both the geometrical and the daylight performance evaluation, it is apparent that the windows' frames extrusion is significant to achieve more accurate results. Geometric accuracy was evaluated using Chamfer and Hausdorff distances, showing good overall accuracy. However, errors were observed in wall heights when the ceiling was not clearly visible and in the separation of windows located close to each other. The models were used for point-in-time grid-based illuminance and view-based luminance simulations in Grasshopper using Honeybee.
For the illuminance simulations, the MAE is approximately 269 lux and the MAPE is 19.5%. For DGP, the MAPE is 7.6% for the RoomPlan models with extruded window frames, with only one misclassification of the DGP category. The results indicate that RoomPlan can be used for visual comfort studies but not for daylight availability studies. Despite these results, suggestions for further work are given, considering both the geometrical and the daylight simulation performance evaluation of the RoomPlan models. ...
The manual models were used as ground truth. For each room, a geometric evaluation and a daylight simulation evaluation were performed using three model versions: manual, RoomPlan with extruded window frames, and RoomPlan without extruded window frames. For both the geometrical and the daylight performance evaluation, it is apparent that the windows' frames extrusion is significant to achieve more accurate results. Geometric accuracy was evaluated using Chamfer and Hausdorff distances, showing good overall accuracy. However, errors were observed in wall heights when the ceiling was not clearly visible and in the separation of windows located close to each other. The models were used for point-in-time grid-based illuminance and view-based luminance simulations in Grasshopper using Honeybee.
For the illuminance simulations, the MAE is approximately 269 lux and the MAPE is 19.5%. For DGP, the MAPE is 7.6% for the RoomPlan models with extruded window frames, with only one misclassification of the DGP category. The results indicate that RoomPlan can be used for visual comfort studies but not for daylight availability studies. Despite these results, suggestions for further work are given, considering both the geometrical and the daylight simulation performance evaluation of the RoomPlan models. ...
This project evaluates the suitability of 3D interior space models acquired using Apple’s RoomPlan API for daylight simulations. The main contribution is a Python-based tool that converts RoomPlan output into HoneyBeeJSON by automatically reconstructing the ceiling and adding window frames, which are missing in RoomPlan’s output. Although RoomPlan is also able to capture furniture, these elements were not used in the geometric evaluation or in the daylight simulations. The resulting models can be directly used in Grasshopper for daylight simulations, reducing the modeling time required by practitioners. To assess the suitability of RoomPlan, three office interiors were scanned using a TLS and modeled both manually and with an iPhone 12 Pro.
The manual models were used as ground truth. For each room, a geometric evaluation and a daylight simulation evaluation were performed using three model versions: manual, RoomPlan with extruded window frames, and RoomPlan without extruded window frames. For both the geometrical and the daylight performance evaluation, it is apparent that the windows' frames extrusion is significant to achieve more accurate results. Geometric accuracy was evaluated using Chamfer and Hausdorff distances, showing good overall accuracy. However, errors were observed in wall heights when the ceiling was not clearly visible and in the separation of windows located close to each other. The models were used for point-in-time grid-based illuminance and view-based luminance simulations in Grasshopper using Honeybee.
For the illuminance simulations, the MAE is approximately 269 lux and the MAPE is 19.5%. For DGP, the MAPE is 7.6% for the RoomPlan models with extruded window frames, with only one misclassification of the DGP category. The results indicate that RoomPlan can be used for visual comfort studies but not for daylight availability studies. Despite these results, suggestions for further work are given, considering both the geometrical and the daylight simulation performance evaluation of the RoomPlan models.
The manual models were used as ground truth. For each room, a geometric evaluation and a daylight simulation evaluation were performed using three model versions: manual, RoomPlan with extruded window frames, and RoomPlan without extruded window frames. For both the geometrical and the daylight performance evaluation, it is apparent that the windows' frames extrusion is significant to achieve more accurate results. Geometric accuracy was evaluated using Chamfer and Hausdorff distances, showing good overall accuracy. However, errors were observed in wall heights when the ceiling was not clearly visible and in the separation of windows located close to each other. The models were used for point-in-time grid-based illuminance and view-based luminance simulations in Grasshopper using Honeybee.
For the illuminance simulations, the MAE is approximately 269 lux and the MAPE is 19.5%. For DGP, the MAPE is 7.6% for the RoomPlan models with extruded window frames, with only one misclassification of the DGP category. The results indicate that RoomPlan can be used for visual comfort studies but not for daylight availability studies. Despite these results, suggestions for further work are given, considering both the geometrical and the daylight simulation performance evaluation of the RoomPlan models.
Structure Guided Roof Heightmap Completion
Via Diffusion Model
Urban digital twins rely on accurate rooftop geometry, yet airborne lidar point clouds are frequently sparse and incomplete, leading to substantial information loss in building reconstruction. This thesis investigates diffusion--based learning as a remedy for high-fidelity roof recovery under severe data corruption.
This thesis proposes a two-stage framework that operates on 2.5D height-map representations. Stage~I introduces a dual-task diffusion model that jointly performs roof height-map completion and roof-line prediction. A novel Bidirectional Control Module enables reciprocal conditioning between the two tasks, enforcing geometric consistency during the denoising process. Stage~II employs a patch-based diffusion upsampler equipped with positional embeddings and a domain-specific global context encoder to synthesise high-resolution height maps while remaining computationally tractable for large and variably-sized buildings. A rigorous preprocessing pipeline further yields two challenging benchmarks, \textsc{S80\_i30} and \textsc{S80\_i80}, derived from 160k real-world building samples.
Extensive experiments conducted on these datasets demonstrate the effectiveness of the proposed approach. Under moderate corruption (\textsc{S80\_i30}), the completion model attains an \textit{RMSE} of \textbf{0.89}~m and a Chamfer distance of \textbf{0.06}, improving upon the state-of-the-art RoofDiffusion baseline by 13.2\% and 17.3\%, respectively. In the severe setting (\textsc{S80\_i80}), the method sustains a 13.5\% \textit{RMSE} reduction. The upsampling stage delivers an additional 10\% \textit{RMSE} gain over the best classical interpolator, and the end-to-end pipeline achieves \textit{RMSE} values of 0.91~m (moderate) and 1.42~m (severe).
The thesis contributes: (i) a structurally-aware diffusion framework for roof completion, (ii) a scalable patch-based upsampler, and (iii) public benchmarks that reflect real lidar degradation. Collectively, these advances close a critical gap between theoretical research and practical generation of LOD2.2 building models, facilitating more reliable urban analytics and planning applications.
...
This thesis proposes a two-stage framework that operates on 2.5D height-map representations. Stage~I introduces a dual-task diffusion model that jointly performs roof height-map completion and roof-line prediction. A novel Bidirectional Control Module enables reciprocal conditioning between the two tasks, enforcing geometric consistency during the denoising process. Stage~II employs a patch-based diffusion upsampler equipped with positional embeddings and a domain-specific global context encoder to synthesise high-resolution height maps while remaining computationally tractable for large and variably-sized buildings. A rigorous preprocessing pipeline further yields two challenging benchmarks, \textsc{S80\_i30} and \textsc{S80\_i80}, derived from 160k real-world building samples.
Extensive experiments conducted on these datasets demonstrate the effectiveness of the proposed approach. Under moderate corruption (\textsc{S80\_i30}), the completion model attains an \textit{RMSE} of \textbf{0.89}~m and a Chamfer distance of \textbf{0.06}, improving upon the state-of-the-art RoofDiffusion baseline by 13.2\% and 17.3\%, respectively. In the severe setting (\textsc{S80\_i80}), the method sustains a 13.5\% \textit{RMSE} reduction. The upsampling stage delivers an additional 10\% \textit{RMSE} gain over the best classical interpolator, and the end-to-end pipeline achieves \textit{RMSE} values of 0.91~m (moderate) and 1.42~m (severe).
The thesis contributes: (i) a structurally-aware diffusion framework for roof completion, (ii) a scalable patch-based upsampler, and (iii) public benchmarks that reflect real lidar degradation. Collectively, these advances close a critical gap between theoretical research and practical generation of LOD2.2 building models, facilitating more reliable urban analytics and planning applications.
...
Urban digital twins rely on accurate rooftop geometry, yet airborne lidar point clouds are frequently sparse and incomplete, leading to substantial information loss in building reconstruction. This thesis investigates diffusion--based learning as a remedy for high-fidelity roof recovery under severe data corruption.
This thesis proposes a two-stage framework that operates on 2.5D height-map representations. Stage~I introduces a dual-task diffusion model that jointly performs roof height-map completion and roof-line prediction. A novel Bidirectional Control Module enables reciprocal conditioning between the two tasks, enforcing geometric consistency during the denoising process. Stage~II employs a patch-based diffusion upsampler equipped with positional embeddings and a domain-specific global context encoder to synthesise high-resolution height maps while remaining computationally tractable for large and variably-sized buildings. A rigorous preprocessing pipeline further yields two challenging benchmarks, \textsc{S80\_i30} and \textsc{S80\_i80}, derived from 160k real-world building samples.
Extensive experiments conducted on these datasets demonstrate the effectiveness of the proposed approach. Under moderate corruption (\textsc{S80\_i30}), the completion model attains an \textit{RMSE} of \textbf{0.89}~m and a Chamfer distance of \textbf{0.06}, improving upon the state-of-the-art RoofDiffusion baseline by 13.2\% and 17.3\%, respectively. In the severe setting (\textsc{S80\_i80}), the method sustains a 13.5\% \textit{RMSE} reduction. The upsampling stage delivers an additional 10\% \textit{RMSE} gain over the best classical interpolator, and the end-to-end pipeline achieves \textit{RMSE} values of 0.91~m (moderate) and 1.42~m (severe).
The thesis contributes: (i) a structurally-aware diffusion framework for roof completion, (ii) a scalable patch-based upsampler, and (iii) public benchmarks that reflect real lidar degradation. Collectively, these advances close a critical gap between theoretical research and practical generation of LOD2.2 building models, facilitating more reliable urban analytics and planning applications.
This thesis proposes a two-stage framework that operates on 2.5D height-map representations. Stage~I introduces a dual-task diffusion model that jointly performs roof height-map completion and roof-line prediction. A novel Bidirectional Control Module enables reciprocal conditioning between the two tasks, enforcing geometric consistency during the denoising process. Stage~II employs a patch-based diffusion upsampler equipped with positional embeddings and a domain-specific global context encoder to synthesise high-resolution height maps while remaining computationally tractable for large and variably-sized buildings. A rigorous preprocessing pipeline further yields two challenging benchmarks, \textsc{S80\_i30} and \textsc{S80\_i80}, derived from 160k real-world building samples.
Extensive experiments conducted on these datasets demonstrate the effectiveness of the proposed approach. Under moderate corruption (\textsc{S80\_i30}), the completion model attains an \textit{RMSE} of \textbf{0.89}~m and a Chamfer distance of \textbf{0.06}, improving upon the state-of-the-art RoofDiffusion baseline by 13.2\% and 17.3\%, respectively. In the severe setting (\textsc{S80\_i80}), the method sustains a 13.5\% \textit{RMSE} reduction. The upsampling stage delivers an additional 10\% \textit{RMSE} gain over the best classical interpolator, and the end-to-end pipeline achieves \textit{RMSE} values of 0.91~m (moderate) and 1.42~m (severe).
The thesis contributes: (i) a structurally-aware diffusion framework for roof completion, (ii) a scalable patch-based upsampler, and (iii) public benchmarks that reflect real lidar degradation. Collectively, these advances close a critical gap between theoretical research and practical generation of LOD2.2 building models, facilitating more reliable urban analytics and planning applications.
High-resolution image mosaicking plays a critical role in geomatics and remote sensing applications, allowing efficient visualization, measurement, and analysis of large-scale envi ronments. Although existing commercial tools provide standard stitching capabilities, they often lack mathematical transparency and real-time customization, limiting their utility in research and professional analysis.
This thesis introduces a systematic approach to dynamic image stitching and visualization within a C# environment. The method uses homography transformations to achieve ac curate image alignment while integrating an optimal seam-finding algorithm to improve visual coherence in overlapping regions. An exportable homography matrix supports co ordinate traceability, enabling users to perform metric evaluations on stitched images. The implementation focuses on creating a lightweight, interactive stitching prototype capable of processing two to three aerial images with high geometric fidelity and run-time efficiency.
Experimental validation confirms that the system delivers precise stitching results and sup ports visual exploration for measurement tasks. By combining mathematical clarity, dy namic responsiveness, and user adaptability, this research contributes to a modular and extensible foundation for image mosaicking in the context of geomatics, with practical rele vance for aerial inspection, photogrammetry, and spatial data visualization ...
This thesis introduces a systematic approach to dynamic image stitching and visualization within a C# environment. The method uses homography transformations to achieve ac curate image alignment while integrating an optimal seam-finding algorithm to improve visual coherence in overlapping regions. An exportable homography matrix supports co ordinate traceability, enabling users to perform metric evaluations on stitched images. The implementation focuses on creating a lightweight, interactive stitching prototype capable of processing two to three aerial images with high geometric fidelity and run-time efficiency.
Experimental validation confirms that the system delivers precise stitching results and sup ports visual exploration for measurement tasks. By combining mathematical clarity, dy namic responsiveness, and user adaptability, this research contributes to a modular and extensible foundation for image mosaicking in the context of geomatics, with practical rele vance for aerial inspection, photogrammetry, and spatial data visualization ...
High-resolution image mosaicking plays a critical role in geomatics and remote sensing applications, allowing efficient visualization, measurement, and analysis of large-scale envi ronments. Although existing commercial tools provide standard stitching capabilities, they often lack mathematical transparency and real-time customization, limiting their utility in research and professional analysis.
This thesis introduces a systematic approach to dynamic image stitching and visualization within a C# environment. The method uses homography transformations to achieve ac curate image alignment while integrating an optimal seam-finding algorithm to improve visual coherence in overlapping regions. An exportable homography matrix supports co ordinate traceability, enabling users to perform metric evaluations on stitched images. The implementation focuses on creating a lightweight, interactive stitching prototype capable of processing two to three aerial images with high geometric fidelity and run-time efficiency.
Experimental validation confirms that the system delivers precise stitching results and sup ports visual exploration for measurement tasks. By combining mathematical clarity, dy namic responsiveness, and user adaptability, this research contributes to a modular and extensible foundation for image mosaicking in the context of geomatics, with practical rele vance for aerial inspection, photogrammetry, and spatial data visualization
This thesis introduces a systematic approach to dynamic image stitching and visualization within a C# environment. The method uses homography transformations to achieve ac curate image alignment while integrating an optimal seam-finding algorithm to improve visual coherence in overlapping regions. An exportable homography matrix supports co ordinate traceability, enabling users to perform metric evaluations on stitched images. The implementation focuses on creating a lightweight, interactive stitching prototype capable of processing two to three aerial images with high geometric fidelity and run-time efficiency.
Experimental validation confirms that the system delivers precise stitching results and sup ports visual exploration for measurement tasks. By combining mathematical clarity, dy namic responsiveness, and user adaptability, this research contributes to a modular and extensible foundation for image mosaicking in the context of geomatics, with practical rele vance for aerial inspection, photogrammetry, and spatial data visualization
This thesis presents a method for extracting structured roof surfaces from remote sensing images. It achieved this by combining semantic segmentation with polygon-based refinement, which allows rooftop boundaries to be described more accurately using line and shape information. The method includes three main stages: (1) using an instance segmentation model to detect and classify rooftop areas; (2) generating polygonal candidates for plannar roof regions based on detected line features; and (3) optimizing label assignments through a Markov Random Field (MRF) model, which integrates prediction confidence with the spatial relationships between polygons. Experiments on benchmark datasets show that this approach improves the accuracy and consistency of rooftop segmentation while reducing incorrect detections. The system is modular and flexible, making it suitable for applications that require reliable roof structure analysis in urban environments.
...
This thesis presents a method for extracting structured roof surfaces from remote sensing images. It achieved this by combining semantic segmentation with polygon-based refinement, which allows rooftop boundaries to be described more accurately using line and shape information. The method includes three main stages: (1) using an instance segmentation model to detect and classify rooftop areas; (2) generating polygonal candidates for plannar roof regions based on detected line features; and (3) optimizing label assignments through a Markov Random Field (MRF) model, which integrates prediction confidence with the spatial relationships between polygons. Experiments on benchmark datasets show that this approach improves the accuracy and consistency of rooftop segmentation while reducing incorrect detections. The system is modular and flexible, making it suitable for applications that require reliable roof structure analysis in urban environments.
Urban Tree Classification in Delft, the Netherlands
Classifying Urban Tree Characteristics with Machine Learning Using Airborne LiDAR and Satellite Imagery
Current urban tree inventories rely heavily on time-consuming manual work and often fail to capture all trees. To effectively monitor the impact of urban trees on their environment and vice versa, an automated method for detecting and grouping trees based on their characteristics is crucial. This research aims to expand current urban tree inventories and cluster trees based on their characteristics. Existing inventories typically include species, age, and height, but trees of the same species and age can vary significantly due to environmental factors. This study focuses on a 500x600 meter area in Delft, encompassing 641 recorded trees from the municipal inventory. An automatic tree detection method using airborne LiDAR (AHN4) point cloud data combined with Random Forest classification was implemented, achieving an accuracy of 85-90%. Individual trees were identified using an existing tree segmentation algorithm, detecting 70% of recorded trees with a mean location difference of 0.71 meters and identifying an additional 460 trees, including those on private land. Despite promising results, limitations include the undetection of small trees (below 3 meters) and classification errors leading to missed detections, multiple identifications for single trees, and false positives. Geometric and reflectance features were extracted. Highresolution, 30cm, spectral images from the SuperView Neo satellites, acquired across three seasons, provided spectral features like tree color and NDVI. Overall resulting in a total of 50 features. This comprehensive inventory allows for clustering of individual trees based on geometric, reflectance, and spectral similarities using a K-means algorithm. The approach enhances urban tree inventories by incorporating new features from airborne Li- DAR and spectral images, such as tree height distribution, crown sphericity, and density. Seasonal changes from spectral images provide insights into tree behavior. These detailed features significantly improve clustering, effectively grouping similar trees together. The findings reveal that trees of the same species in seemingly similar environments exhibit significantly different characteristics. This research offers a method to enhance urban tree inventories and supports long-term studies to reveal how trees respond to urban development, climate change, and ecological dynamics. Correlating tree growth and health with specific locations and climatic conditions can aid in developing sustainable urban planning and conservation strategies.
...
Current urban tree inventories rely heavily on time-consuming manual work and often fail to capture all trees. To effectively monitor the impact of urban trees on their environment and vice versa, an automated method for detecting and grouping trees based on their characteristics is crucial. This research aims to expand current urban tree inventories and cluster trees based on their characteristics. Existing inventories typically include species, age, and height, but trees of the same species and age can vary significantly due to environmental factors. This study focuses on a 500x600 meter area in Delft, encompassing 641 recorded trees from the municipal inventory. An automatic tree detection method using airborne LiDAR (AHN4) point cloud data combined with Random Forest classification was implemented, achieving an accuracy of 85-90%. Individual trees were identified using an existing tree segmentation algorithm, detecting 70% of recorded trees with a mean location difference of 0.71 meters and identifying an additional 460 trees, including those on private land. Despite promising results, limitations include the undetection of small trees (below 3 meters) and classification errors leading to missed detections, multiple identifications for single trees, and false positives. Geometric and reflectance features were extracted. Highresolution, 30cm, spectral images from the SuperView Neo satellites, acquired across three seasons, provided spectral features like tree color and NDVI. Overall resulting in a total of 50 features. This comprehensive inventory allows for clustering of individual trees based on geometric, reflectance, and spectral similarities using a K-means algorithm. The approach enhances urban tree inventories by incorporating new features from airborne Li- DAR and spectral images, such as tree height distribution, crown sphericity, and density. Seasonal changes from spectral images provide insights into tree behavior. These detailed features significantly improve clustering, effectively grouping similar trees together. The findings reveal that trees of the same species in seemingly similar environments exhibit significantly different characteristics. This research offers a method to enhance urban tree inventories and supports long-term studies to reveal how trees respond to urban development, climate change, and ecological dynamics. Correlating tree growth and health with specific locations and climatic conditions can aid in developing sustainable urban planning and conservation strategies.
The accuracy and comprehensiveness of 3D city models have become increasingly important for applications like monitoring, sustainability evaluation, disaster management, and urban planning. However, creating accurate and complete 3D models is challenging. Traditional methods, such as photogrammetry and Light Detection and Ranging (LiDAR), often face issues like time-consuming processes and data gaps caused by occlusion. Additionally, these methods typically produce discrete models that fail to capture all the information from the original objects, limiting the ability to reconstruct small structures.
In this thesis, we explore the potential and characteristics of enhancing 3D models for urban areas using implicit neural representation. This method offers generalizability, ensures no void areas, and provides more detailed information when using multi-modal inputs. The datasets used for model training include Actueel Hoogtebestand Nederland 3 (AHN3), 2019 Luchtfoto Beeldmateriaal, 3D Basisregistratie Adressen en Gebouwen (BAG), and Basisregistratie Grootschalige Topografie (BGT). The city center of Eindhoven is used for training and testing, and the city center of Rotterdam serves as a test dataset.
The method involves learning location-dependent latent codes from raw point cloud and orthophoto, and adjusting the probability of space occupancy based on point clouds sampled from the 3D city model. A decoder is used to calculate the probability of existence for any point in the 3D space from the continuous field. Proper sampling allows us to derive a continuous Digital Surface Model (DSM) with unlimited resolution from the neural network.
The results show a high degree of accuracy: the generated DSM for the training area in Eindhoven demonstrated a median absolute error of 0.484 m overall. Accuracy for building areas was recorded at 0.798 m, and for terrain, it was 0.302 m. The model also displayed robust generalization capabilities, with an accuracy of 0.336 m in the Eindhoven test area and 0.23 meters in Rotterdam. Additionally, the model's ability to fill voids was confirmed through both visual inspection and quantitative evaluations. This research underscores the potential of implicit neural representation for generating detailed DSMs and effectively filling no-data voids. ...
In this thesis, we explore the potential and characteristics of enhancing 3D models for urban areas using implicit neural representation. This method offers generalizability, ensures no void areas, and provides more detailed information when using multi-modal inputs. The datasets used for model training include Actueel Hoogtebestand Nederland 3 (AHN3), 2019 Luchtfoto Beeldmateriaal, 3D Basisregistratie Adressen en Gebouwen (BAG), and Basisregistratie Grootschalige Topografie (BGT). The city center of Eindhoven is used for training and testing, and the city center of Rotterdam serves as a test dataset.
The method involves learning location-dependent latent codes from raw point cloud and orthophoto, and adjusting the probability of space occupancy based on point clouds sampled from the 3D city model. A decoder is used to calculate the probability of existence for any point in the 3D space from the continuous field. Proper sampling allows us to derive a continuous Digital Surface Model (DSM) with unlimited resolution from the neural network.
The results show a high degree of accuracy: the generated DSM for the training area in Eindhoven demonstrated a median absolute error of 0.484 m overall. Accuracy for building areas was recorded at 0.798 m, and for terrain, it was 0.302 m. The model also displayed robust generalization capabilities, with an accuracy of 0.336 m in the Eindhoven test area and 0.23 meters in Rotterdam. Additionally, the model's ability to fill voids was confirmed through both visual inspection and quantitative evaluations. This research underscores the potential of implicit neural representation for generating detailed DSMs and effectively filling no-data voids. ...
The accuracy and comprehensiveness of 3D city models have become increasingly important for applications like monitoring, sustainability evaluation, disaster management, and urban planning. However, creating accurate and complete 3D models is challenging. Traditional methods, such as photogrammetry and Light Detection and Ranging (LiDAR), often face issues like time-consuming processes and data gaps caused by occlusion. Additionally, these methods typically produce discrete models that fail to capture all the information from the original objects, limiting the ability to reconstruct small structures.
In this thesis, we explore the potential and characteristics of enhancing 3D models for urban areas using implicit neural representation. This method offers generalizability, ensures no void areas, and provides more detailed information when using multi-modal inputs. The datasets used for model training include Actueel Hoogtebestand Nederland 3 (AHN3), 2019 Luchtfoto Beeldmateriaal, 3D Basisregistratie Adressen en Gebouwen (BAG), and Basisregistratie Grootschalige Topografie (BGT). The city center of Eindhoven is used for training and testing, and the city center of Rotterdam serves as a test dataset.
The method involves learning location-dependent latent codes from raw point cloud and orthophoto, and adjusting the probability of space occupancy based on point clouds sampled from the 3D city model. A decoder is used to calculate the probability of existence for any point in the 3D space from the continuous field. Proper sampling allows us to derive a continuous Digital Surface Model (DSM) with unlimited resolution from the neural network.
The results show a high degree of accuracy: the generated DSM for the training area in Eindhoven demonstrated a median absolute error of 0.484 m overall. Accuracy for building areas was recorded at 0.798 m, and for terrain, it was 0.302 m. The model also displayed robust generalization capabilities, with an accuracy of 0.336 m in the Eindhoven test area and 0.23 meters in Rotterdam. Additionally, the model's ability to fill voids was confirmed through both visual inspection and quantitative evaluations. This research underscores the potential of implicit neural representation for generating detailed DSMs and effectively filling no-data voids.
In this thesis, we explore the potential and characteristics of enhancing 3D models for urban areas using implicit neural representation. This method offers generalizability, ensures no void areas, and provides more detailed information when using multi-modal inputs. The datasets used for model training include Actueel Hoogtebestand Nederland 3 (AHN3), 2019 Luchtfoto Beeldmateriaal, 3D Basisregistratie Adressen en Gebouwen (BAG), and Basisregistratie Grootschalige Topografie (BGT). The city center of Eindhoven is used for training and testing, and the city center of Rotterdam serves as a test dataset.
The method involves learning location-dependent latent codes from raw point cloud and orthophoto, and adjusting the probability of space occupancy based on point clouds sampled from the 3D city model. A decoder is used to calculate the probability of existence for any point in the 3D space from the continuous field. Proper sampling allows us to derive a continuous Digital Surface Model (DSM) with unlimited resolution from the neural network.
The results show a high degree of accuracy: the generated DSM for the training area in Eindhoven demonstrated a median absolute error of 0.484 m overall. Accuracy for building areas was recorded at 0.798 m, and for terrain, it was 0.302 m. The model also displayed robust generalization capabilities, with an accuracy of 0.336 m in the Eindhoven test area and 0.23 meters in Rotterdam. Additionally, the model's ability to fill voids was confirmed through both visual inspection and quantitative evaluations. This research underscores the potential of implicit neural representation for generating detailed DSMs and effectively filling no-data voids.
In response to global climate change and the increasing frequency of extreme natural hazards, an early intervention flood risk assessment methodology has been developed to assess damage at the building component level. This approach allows for the precise identification of vulnerable structural components and localized redesigns, leading to more cost-effective and safer structural solutions. The focus of this paper is on quantifying flood damage to components and visualizing the results, with two main objectives: 1) quantifying the damage and 2) interactive visualization.
Hydrodynamic simulations of compound floods under future climate scenarios are performed to apply the quantitative results at the micro level. Using future climate scenarios obtained by perturbing atmospheric data with a regional climate model, the top 24 hours of flood-driven data are selected to create inundation maps. Vulnerability curves for each material are plotted, taking into account the sensitivity of the component materials to water. Based on the water depth at each point, the failure probability is determined. This method of quantifying component vulnerability provides new insights into micro-level flood damage.
Finally, an integrated BIM-GIS model is constructed, and the damage level and distribution are displayed on the ArcGIS Pro platform. This allows different stakeholders to intuitively obtain damage level information. Users can customize their view, for example, by focusing solely on the damage to first-floor exterior windows or by examining the damage levels of components and buildings under different intensity flood ratings. ...
Hydrodynamic simulations of compound floods under future climate scenarios are performed to apply the quantitative results at the micro level. Using future climate scenarios obtained by perturbing atmospheric data with a regional climate model, the top 24 hours of flood-driven data are selected to create inundation maps. Vulnerability curves for each material are plotted, taking into account the sensitivity of the component materials to water. Based on the water depth at each point, the failure probability is determined. This method of quantifying component vulnerability provides new insights into micro-level flood damage.
Finally, an integrated BIM-GIS model is constructed, and the damage level and distribution are displayed on the ArcGIS Pro platform. This allows different stakeholders to intuitively obtain damage level information. Users can customize their view, for example, by focusing solely on the damage to first-floor exterior windows or by examining the damage levels of components and buildings under different intensity flood ratings. ...
In response to global climate change and the increasing frequency of extreme natural hazards, an early intervention flood risk assessment methodology has been developed to assess damage at the building component level. This approach allows for the precise identification of vulnerable structural components and localized redesigns, leading to more cost-effective and safer structural solutions. The focus of this paper is on quantifying flood damage to components and visualizing the results, with two main objectives: 1) quantifying the damage and 2) interactive visualization.
Hydrodynamic simulations of compound floods under future climate scenarios are performed to apply the quantitative results at the micro level. Using future climate scenarios obtained by perturbing atmospheric data with a regional climate model, the top 24 hours of flood-driven data are selected to create inundation maps. Vulnerability curves for each material are plotted, taking into account the sensitivity of the component materials to water. Based on the water depth at each point, the failure probability is determined. This method of quantifying component vulnerability provides new insights into micro-level flood damage.
Finally, an integrated BIM-GIS model is constructed, and the damage level and distribution are displayed on the ArcGIS Pro platform. This allows different stakeholders to intuitively obtain damage level information. Users can customize their view, for example, by focusing solely on the damage to first-floor exterior windows or by examining the damage levels of components and buildings under different intensity flood ratings.
Hydrodynamic simulations of compound floods under future climate scenarios are performed to apply the quantitative results at the micro level. Using future climate scenarios obtained by perturbing atmospheric data with a regional climate model, the top 24 hours of flood-driven data are selected to create inundation maps. Vulnerability curves for each material are plotted, taking into account the sensitivity of the component materials to water. Based on the water depth at each point, the failure probability is determined. This method of quantifying component vulnerability provides new insights into micro-level flood damage.
Finally, an integrated BIM-GIS model is constructed, and the damage level and distribution are displayed on the ArcGIS Pro platform. This allows different stakeholders to intuitively obtain damage level information. Users can customize their view, for example, by focusing solely on the damage to first-floor exterior windows or by examining the damage levels of components and buildings under different intensity flood ratings.
Integrating urban context in daylighting simulation
The design consequences in Dutch urban areas, regarding visual & non-visual levels of daylight
The Netherlands is facing a housing demand of 1 million homes before 2030. Most of these residences are planned to be built in and around existing cities, causing an increase in urban densities with sub-optimal indoor daylighting conditions as a result. Simultaneously, the daylight assessment methodology for buildings in the Netherlands is set to change from the Dutch NEN 2057 to the European EN 17037. The European norm uses more accurate metrics to express daylighting performance but does not consider urban context (i.e. external buildings) in the simulation models. As a result, a concern is that indoor daylighting in dense urban areas is inadequately protected. Moreover, it is unknown to what extent the urban context affects the well-being of humans, regarding visual and non-visual levels of daylight.
A multitude of daylight simulations is run and analysed in the thesis to better understand the impact of the urban context on indoor daylighting performance. Visual daylighting is assessed following the EN 17037 methodology with urban context integrated. Non-visual daylight performance is assessed using two novel metrics: melanopic autonomy and melanopic isotropy. The results have revealed that the discrepancy between simulations with and without the integration of urban context is up to 90% for realistic residences throughout the Netherlands, depending on urban characteristics and density. On average, indoor daylighting is decreased by 36% when the urban context is integrated with the EN 17037. The non-visual stimulus was found to be sufficient in residences that are compliant with EUmin levels but insufficient for residences that only comply with the Dutch building code. Sky view factor (SVF) and Building Floor were found to be useful indicators of daylighting performance in early design stages. Urban density indicators such as the FSI and OSR seem to be negatively correlated with daylighting performance.
The thesis concludes with the advice to include urban context in daylighting simulations so that bad daylighting can be properly mitigated. Effective mitigation strategies are increasing glass transmission values, interior reflectance values, and exterior building reflectance values. Another effective strategy is to avoid bad daylighting conditions in the first place by not positioning residences on the first 5 building floors in high-density urban areas. The results from this thesis can be used by daylighting designers and architects who are interested in ensuring adequate and healthy daylighting conditions in the residences they design: not only in digital environments but in the real world. ...
A multitude of daylight simulations is run and analysed in the thesis to better understand the impact of the urban context on indoor daylighting performance. Visual daylighting is assessed following the EN 17037 methodology with urban context integrated. Non-visual daylight performance is assessed using two novel metrics: melanopic autonomy and melanopic isotropy. The results have revealed that the discrepancy between simulations with and without the integration of urban context is up to 90% for realistic residences throughout the Netherlands, depending on urban characteristics and density. On average, indoor daylighting is decreased by 36% when the urban context is integrated with the EN 17037. The non-visual stimulus was found to be sufficient in residences that are compliant with EUmin levels but insufficient for residences that only comply with the Dutch building code. Sky view factor (SVF) and Building Floor were found to be useful indicators of daylighting performance in early design stages. Urban density indicators such as the FSI and OSR seem to be negatively correlated with daylighting performance.
The thesis concludes with the advice to include urban context in daylighting simulations so that bad daylighting can be properly mitigated. Effective mitigation strategies are increasing glass transmission values, interior reflectance values, and exterior building reflectance values. Another effective strategy is to avoid bad daylighting conditions in the first place by not positioning residences on the first 5 building floors in high-density urban areas. The results from this thesis can be used by daylighting designers and architects who are interested in ensuring adequate and healthy daylighting conditions in the residences they design: not only in digital environments but in the real world. ...
The Netherlands is facing a housing demand of 1 million homes before 2030. Most of these residences are planned to be built in and around existing cities, causing an increase in urban densities with sub-optimal indoor daylighting conditions as a result. Simultaneously, the daylight assessment methodology for buildings in the Netherlands is set to change from the Dutch NEN 2057 to the European EN 17037. The European norm uses more accurate metrics to express daylighting performance but does not consider urban context (i.e. external buildings) in the simulation models. As a result, a concern is that indoor daylighting in dense urban areas is inadequately protected. Moreover, it is unknown to what extent the urban context affects the well-being of humans, regarding visual and non-visual levels of daylight.
A multitude of daylight simulations is run and analysed in the thesis to better understand the impact of the urban context on indoor daylighting performance. Visual daylighting is assessed following the EN 17037 methodology with urban context integrated. Non-visual daylight performance is assessed using two novel metrics: melanopic autonomy and melanopic isotropy. The results have revealed that the discrepancy between simulations with and without the integration of urban context is up to 90% for realistic residences throughout the Netherlands, depending on urban characteristics and density. On average, indoor daylighting is decreased by 36% when the urban context is integrated with the EN 17037. The non-visual stimulus was found to be sufficient in residences that are compliant with EUmin levels but insufficient for residences that only comply with the Dutch building code. Sky view factor (SVF) and Building Floor were found to be useful indicators of daylighting performance in early design stages. Urban density indicators such as the FSI and OSR seem to be negatively correlated with daylighting performance.
The thesis concludes with the advice to include urban context in daylighting simulations so that bad daylighting can be properly mitigated. Effective mitigation strategies are increasing glass transmission values, interior reflectance values, and exterior building reflectance values. Another effective strategy is to avoid bad daylighting conditions in the first place by not positioning residences on the first 5 building floors in high-density urban areas. The results from this thesis can be used by daylighting designers and architects who are interested in ensuring adequate and healthy daylighting conditions in the residences they design: not only in digital environments but in the real world.
A multitude of daylight simulations is run and analysed in the thesis to better understand the impact of the urban context on indoor daylighting performance. Visual daylighting is assessed following the EN 17037 methodology with urban context integrated. Non-visual daylight performance is assessed using two novel metrics: melanopic autonomy and melanopic isotropy. The results have revealed that the discrepancy between simulations with and without the integration of urban context is up to 90% for realistic residences throughout the Netherlands, depending on urban characteristics and density. On average, indoor daylighting is decreased by 36% when the urban context is integrated with the EN 17037. The non-visual stimulus was found to be sufficient in residences that are compliant with EUmin levels but insufficient for residences that only comply with the Dutch building code. Sky view factor (SVF) and Building Floor were found to be useful indicators of daylighting performance in early design stages. Urban density indicators such as the FSI and OSR seem to be negatively correlated with daylighting performance.
The thesis concludes with the advice to include urban context in daylighting simulations so that bad daylighting can be properly mitigated. Effective mitigation strategies are increasing glass transmission values, interior reflectance values, and exterior building reflectance values. Another effective strategy is to avoid bad daylighting conditions in the first place by not positioning residences on the first 5 building floors in high-density urban areas. The results from this thesis can be used by daylighting designers and architects who are interested in ensuring adequate and healthy daylighting conditions in the residences they design: not only in digital environments but in the real world.
Digital Design Tool For Climate Resilient Buildings
Designing an open-source Python tool to assess the climate resilience of structural IFC models regarding Climate Change in the Netherlands
Due to global warming, the Netherlands is experiencing a variety of climatic changes, including temperature rise, increased solar radiation and condensation, low pressure, high humidity, wildfires, drought, subsidence, changes in groundwater levels, an increased risk of flooding, and downbursts, thunder, wind gusts, and hail. Building materials, including steel, concrete, and timber, are affected directly or indirectly by these climatic events.
When it comes to resilience, increasing moisture, temperature, subsidence, and flood damage affect structural materials most. Investigating the impact of temperature and flood damage on construction materials was the main goal of the thesis.
For flood assessment, flood loads based on the FEMA Coastal Construction Handbook are used to simulate flooding damage and evaluate the impact on structures. Hand calculations are used to calculate the deflection. Damage evaluation involves calculations from reference cases and utilizes databases such as Hazus, the REDi rating system, and FEMA to determine recovery and repair times. The script for the assessment tool incorporates these calculations, formulas, and numbers, eventually resulting in a graph based on the dimension of the column and the material; the deflection, damage, recovery time, and repair time are returned. The current flood assessment is limited to deflection in terms of structural assessment, but it can easily be expanded to contain stress calculations or other similar formulas.
Regarding the temperature impact, an empirical concrete corrosion formula calculates mass loss, while the Arrhenius equation assesses the deterioration of wood and steel. A specific formula for concrete corrosion considering the concrete layer is required. Using Faraday's equation, corrosion ratios or material degradation ratios can be converted into mm/year, determining the new cross-section size and assessing its impact on structural deflection by anticipating the mass loss. This loss is also directly linked to the structure's performance in a flood, effectively integrating both investigated events.
Both flood assessment and temperature effect approaches are translated into a Python script using packages like open-meteo for climate data, klimaateffectatlas for flood depths, and ifcopenshell for data extraction from IFC models. ...
When it comes to resilience, increasing moisture, temperature, subsidence, and flood damage affect structural materials most. Investigating the impact of temperature and flood damage on construction materials was the main goal of the thesis.
For flood assessment, flood loads based on the FEMA Coastal Construction Handbook are used to simulate flooding damage and evaluate the impact on structures. Hand calculations are used to calculate the deflection. Damage evaluation involves calculations from reference cases and utilizes databases such as Hazus, the REDi rating system, and FEMA to determine recovery and repair times. The script for the assessment tool incorporates these calculations, formulas, and numbers, eventually resulting in a graph based on the dimension of the column and the material; the deflection, damage, recovery time, and repair time are returned. The current flood assessment is limited to deflection in terms of structural assessment, but it can easily be expanded to contain stress calculations or other similar formulas.
Regarding the temperature impact, an empirical concrete corrosion formula calculates mass loss, while the Arrhenius equation assesses the deterioration of wood and steel. A specific formula for concrete corrosion considering the concrete layer is required. Using Faraday's equation, corrosion ratios or material degradation ratios can be converted into mm/year, determining the new cross-section size and assessing its impact on structural deflection by anticipating the mass loss. This loss is also directly linked to the structure's performance in a flood, effectively integrating both investigated events.
Both flood assessment and temperature effect approaches are translated into a Python script using packages like open-meteo for climate data, klimaateffectatlas for flood depths, and ifcopenshell for data extraction from IFC models. ...
Due to global warming, the Netherlands is experiencing a variety of climatic changes, including temperature rise, increased solar radiation and condensation, low pressure, high humidity, wildfires, drought, subsidence, changes in groundwater levels, an increased risk of flooding, and downbursts, thunder, wind gusts, and hail. Building materials, including steel, concrete, and timber, are affected directly or indirectly by these climatic events.
When it comes to resilience, increasing moisture, temperature, subsidence, and flood damage affect structural materials most. Investigating the impact of temperature and flood damage on construction materials was the main goal of the thesis.
For flood assessment, flood loads based on the FEMA Coastal Construction Handbook are used to simulate flooding damage and evaluate the impact on structures. Hand calculations are used to calculate the deflection. Damage evaluation involves calculations from reference cases and utilizes databases such as Hazus, the REDi rating system, and FEMA to determine recovery and repair times. The script for the assessment tool incorporates these calculations, formulas, and numbers, eventually resulting in a graph based on the dimension of the column and the material; the deflection, damage, recovery time, and repair time are returned. The current flood assessment is limited to deflection in terms of structural assessment, but it can easily be expanded to contain stress calculations or other similar formulas.
Regarding the temperature impact, an empirical concrete corrosion formula calculates mass loss, while the Arrhenius equation assesses the deterioration of wood and steel. A specific formula for concrete corrosion considering the concrete layer is required. Using Faraday's equation, corrosion ratios or material degradation ratios can be converted into mm/year, determining the new cross-section size and assessing its impact on structural deflection by anticipating the mass loss. This loss is also directly linked to the structure's performance in a flood, effectively integrating both investigated events.
Both flood assessment and temperature effect approaches are translated into a Python script using packages like open-meteo for climate data, klimaateffectatlas for flood depths, and ifcopenshell for data extraction from IFC models.
When it comes to resilience, increasing moisture, temperature, subsidence, and flood damage affect structural materials most. Investigating the impact of temperature and flood damage on construction materials was the main goal of the thesis.
For flood assessment, flood loads based on the FEMA Coastal Construction Handbook are used to simulate flooding damage and evaluate the impact on structures. Hand calculations are used to calculate the deflection. Damage evaluation involves calculations from reference cases and utilizes databases such as Hazus, the REDi rating system, and FEMA to determine recovery and repair times. The script for the assessment tool incorporates these calculations, formulas, and numbers, eventually resulting in a graph based on the dimension of the column and the material; the deflection, damage, recovery time, and repair time are returned. The current flood assessment is limited to deflection in terms of structural assessment, but it can easily be expanded to contain stress calculations or other similar formulas.
Regarding the temperature impact, an empirical concrete corrosion formula calculates mass loss, while the Arrhenius equation assesses the deterioration of wood and steel. A specific formula for concrete corrosion considering the concrete layer is required. Using Faraday's equation, corrosion ratios or material degradation ratios can be converted into mm/year, determining the new cross-section size and assessing its impact on structural deflection by anticipating the mass loss. This loss is also directly linked to the structure's performance in a flood, effectively integrating both investigated events.
Both flood assessment and temperature effect approaches are translated into a Python script using packages like open-meteo for climate data, klimaateffectatlas for flood depths, and ifcopenshell for data extraction from IFC models.
Nowadays, humans rely in technology more and more when it comes to navigation and localisation and in many aspects of life as well. While most concepts related to localisation and navigation of outdoors environments are already well derived from various researches and softwares, the indoor environment remains a significantly unexplored area. Nevertheless, lately there have been increased interest on Location Based Services (LBS) and Indoor Positioning Systems (IPS). There are already several methods available for indoor localisation such as Wi-Fi Fingerprinting and Bluetooth Beacons, but none of them is fully functional yet. It remains a field that requires more and further research and investigation in order to reach a satisfactory and complete Indoor Localisation-Navigation method.
Therefore, this thesis's main objective is to investigate and explore a new method for Indoor Localisation based on Isovists. The exploration and evaluation of Isovist-Fingerprinting approach for Indoor Localisation can extend the fields of LBS and Geomatics. The main research question is “To what extent can isovist support Indoor Localisation” and through this and a series of sub-questions to analyse the Isovist concept in relation to the Indoor Localisation. This is achieved by forming a proof of concept and a methodology that investigates how the Isovists would benefit an LBS.
To succeed that the methodology is divided into 4 main sections. The Data Acquisition for which the newly supported from smartphones Light Detection And Ranging (LiDAR) technology were used. The Space Syntax and Isovist Analysis Measures, where all the concepts related such as the Isovist Parameters were analysed in depth for better understanding of their effect. Then the Matching and Localisation Algorithms, where the possibilities and options on how to reach the localisation were investigated and analysed. And finally, the Tests and Experiments took place in order to evaluate all the prior stages of the methodology.
The main conclusion of this research is that a method for Indoor Localisation based on Isovists is feasible and can indeed support an LBS. The analysis and evaluation of all related components has be done and if putting all the parts in the right order they can be of high value for LBS applications. Since is a new method of Indoor Localisation, there is plenty of future work to be done which mainly focuses on how to connect it with existing techniques and integrate all together into a user application. ...
Therefore, this thesis's main objective is to investigate and explore a new method for Indoor Localisation based on Isovists. The exploration and evaluation of Isovist-Fingerprinting approach for Indoor Localisation can extend the fields of LBS and Geomatics. The main research question is “To what extent can isovist support Indoor Localisation” and through this and a series of sub-questions to analyse the Isovist concept in relation to the Indoor Localisation. This is achieved by forming a proof of concept and a methodology that investigates how the Isovists would benefit an LBS.
To succeed that the methodology is divided into 4 main sections. The Data Acquisition for which the newly supported from smartphones Light Detection And Ranging (LiDAR) technology were used. The Space Syntax and Isovist Analysis Measures, where all the concepts related such as the Isovist Parameters were analysed in depth for better understanding of their effect. Then the Matching and Localisation Algorithms, where the possibilities and options on how to reach the localisation were investigated and analysed. And finally, the Tests and Experiments took place in order to evaluate all the prior stages of the methodology.
The main conclusion of this research is that a method for Indoor Localisation based on Isovists is feasible and can indeed support an LBS. The analysis and evaluation of all related components has be done and if putting all the parts in the right order they can be of high value for LBS applications. Since is a new method of Indoor Localisation, there is plenty of future work to be done which mainly focuses on how to connect it with existing techniques and integrate all together into a user application. ...
Nowadays, humans rely in technology more and more when it comes to navigation and localisation and in many aspects of life as well. While most concepts related to localisation and navigation of outdoors environments are already well derived from various researches and softwares, the indoor environment remains a significantly unexplored area. Nevertheless, lately there have been increased interest on Location Based Services (LBS) and Indoor Positioning Systems (IPS). There are already several methods available for indoor localisation such as Wi-Fi Fingerprinting and Bluetooth Beacons, but none of them is fully functional yet. It remains a field that requires more and further research and investigation in order to reach a satisfactory and complete Indoor Localisation-Navigation method.
Therefore, this thesis's main objective is to investigate and explore a new method for Indoor Localisation based on Isovists. The exploration and evaluation of Isovist-Fingerprinting approach for Indoor Localisation can extend the fields of LBS and Geomatics. The main research question is “To what extent can isovist support Indoor Localisation” and through this and a series of sub-questions to analyse the Isovist concept in relation to the Indoor Localisation. This is achieved by forming a proof of concept and a methodology that investigates how the Isovists would benefit an LBS.
To succeed that the methodology is divided into 4 main sections. The Data Acquisition for which the newly supported from smartphones Light Detection And Ranging (LiDAR) technology were used. The Space Syntax and Isovist Analysis Measures, where all the concepts related such as the Isovist Parameters were analysed in depth for better understanding of their effect. Then the Matching and Localisation Algorithms, where the possibilities and options on how to reach the localisation were investigated and analysed. And finally, the Tests and Experiments took place in order to evaluate all the prior stages of the methodology.
The main conclusion of this research is that a method for Indoor Localisation based on Isovists is feasible and can indeed support an LBS. The analysis and evaluation of all related components has be done and if putting all the parts in the right order they can be of high value for LBS applications. Since is a new method of Indoor Localisation, there is plenty of future work to be done which mainly focuses on how to connect it with existing techniques and integrate all together into a user application.
Therefore, this thesis's main objective is to investigate and explore a new method for Indoor Localisation based on Isovists. The exploration and evaluation of Isovist-Fingerprinting approach for Indoor Localisation can extend the fields of LBS and Geomatics. The main research question is “To what extent can isovist support Indoor Localisation” and through this and a series of sub-questions to analyse the Isovist concept in relation to the Indoor Localisation. This is achieved by forming a proof of concept and a methodology that investigates how the Isovists would benefit an LBS.
To succeed that the methodology is divided into 4 main sections. The Data Acquisition for which the newly supported from smartphones Light Detection And Ranging (LiDAR) technology were used. The Space Syntax and Isovist Analysis Measures, where all the concepts related such as the Isovist Parameters were analysed in depth for better understanding of their effect. Then the Matching and Localisation Algorithms, where the possibilities and options on how to reach the localisation were investigated and analysed. And finally, the Tests and Experiments took place in order to evaluate all the prior stages of the methodology.
The main conclusion of this research is that a method for Indoor Localisation based on Isovists is feasible and can indeed support an LBS. The analysis and evaluation of all related components has be done and if putting all the parts in the right order they can be of high value for LBS applications. Since is a new method of Indoor Localisation, there is plenty of future work to be done which mainly focuses on how to connect it with existing techniques and integrate all together into a user application.
Nowadays, the evolution of localisation and navigation technologies is vast, aiding towards facilitating users’ guidance in various environments. Outdoor positioning can be easily achieved, with the widely used Global Navigation Satellite Systems (GNSS), which comprise a universal standard for positioning and are included in every person’s mobile device. However, due to the presence of high buildings in dense urban environments and bad reception in indoor environments, the performance of GNSS is significantly degraded. Therefore, alternative ways of positioning and localisation respectively, need to be explored. In indoor environments, unlike outdoors, there is no universal standard, as the different indoor localisation techniques, that are currently implemented have their own bottlenecks. The most widely used Wi-Fi fingerprinting, requires a constantly up-to-date radio map of the signals from the Wi-Fi access points, whose creation is also a heavy and time-consuming technique. Additionally, other techniques require an installation of costly sensors or either equipment.
Therefore, this thesis investigates the possibility of the ceilings in public or semi-public buildings, being used for indoor localisation, by using features that are included in a simple mobile device. The research additionally involves location tracking of different users, in order to discover different movement patterns in an indoor facility. Indoor localisation is achieved based on the comparison of user and reference data, that can be both point clouds and images, using the Light detection and ranging (LiDAR) of an iPad 12 pro and camera sensors of an Android device. The point cloud-based localisation is implemented based on different combinations of global and local registration techniques, while the image-based approach involves different feature detection, description and matching techniques. Using a web-application to visualise the indoor localisation results, an indoor model and a network graph of the Faculty of Architecture and the Built Environment, location tracking of different users is implemented and visualised in a heat-map. Additionally, a dashboard is created that can be used by a facility manager to translate the user paths to valuable information and reveal different movement patterns in an indoor facility.
The followed methodology showed promising results, concerning the reliability of ceilings for real-time indoor localisation, based on LiDAR and camera sensors, that are incorporated in up-to-date mobile devices. The robustness of Colored Iterative Closest Point (ICP) algorithm for indoor localisation based on point clouds was revealed, both in terms of time efficiency and quality, while the combination of Speeded-Up Robust Features (SURF) feature detector and Scale Invariant Feature Transform (SIFT) descriptor provides the optimal indoor localisation results with image data. The proposed pipeline revealed encouraging results for use in emergency situations, based on static data acquisition of a user, while it is also suitable for dynamic applications, in case a sensor is mounted on an automated device for indoor mapping operations.
...
Therefore, this thesis investigates the possibility of the ceilings in public or semi-public buildings, being used for indoor localisation, by using features that are included in a simple mobile device. The research additionally involves location tracking of different users, in order to discover different movement patterns in an indoor facility. Indoor localisation is achieved based on the comparison of user and reference data, that can be both point clouds and images, using the Light detection and ranging (LiDAR) of an iPad 12 pro and camera sensors of an Android device. The point cloud-based localisation is implemented based on different combinations of global and local registration techniques, while the image-based approach involves different feature detection, description and matching techniques. Using a web-application to visualise the indoor localisation results, an indoor model and a network graph of the Faculty of Architecture and the Built Environment, location tracking of different users is implemented and visualised in a heat-map. Additionally, a dashboard is created that can be used by a facility manager to translate the user paths to valuable information and reveal different movement patterns in an indoor facility.
The followed methodology showed promising results, concerning the reliability of ceilings for real-time indoor localisation, based on LiDAR and camera sensors, that are incorporated in up-to-date mobile devices. The robustness of Colored Iterative Closest Point (ICP) algorithm for indoor localisation based on point clouds was revealed, both in terms of time efficiency and quality, while the combination of Speeded-Up Robust Features (SURF) feature detector and Scale Invariant Feature Transform (SIFT) descriptor provides the optimal indoor localisation results with image data. The proposed pipeline revealed encouraging results for use in emergency situations, based on static data acquisition of a user, while it is also suitable for dynamic applications, in case a sensor is mounted on an automated device for indoor mapping operations.
...
Nowadays, the evolution of localisation and navigation technologies is vast, aiding towards facilitating users’ guidance in various environments. Outdoor positioning can be easily achieved, with the widely used Global Navigation Satellite Systems (GNSS), which comprise a universal standard for positioning and are included in every person’s mobile device. However, due to the presence of high buildings in dense urban environments and bad reception in indoor environments, the performance of GNSS is significantly degraded. Therefore, alternative ways of positioning and localisation respectively, need to be explored. In indoor environments, unlike outdoors, there is no universal standard, as the different indoor localisation techniques, that are currently implemented have their own bottlenecks. The most widely used Wi-Fi fingerprinting, requires a constantly up-to-date radio map of the signals from the Wi-Fi access points, whose creation is also a heavy and time-consuming technique. Additionally, other techniques require an installation of costly sensors or either equipment.
Therefore, this thesis investigates the possibility of the ceilings in public or semi-public buildings, being used for indoor localisation, by using features that are included in a simple mobile device. The research additionally involves location tracking of different users, in order to discover different movement patterns in an indoor facility. Indoor localisation is achieved based on the comparison of user and reference data, that can be both point clouds and images, using the Light detection and ranging (LiDAR) of an iPad 12 pro and camera sensors of an Android device. The point cloud-based localisation is implemented based on different combinations of global and local registration techniques, while the image-based approach involves different feature detection, description and matching techniques. Using a web-application to visualise the indoor localisation results, an indoor model and a network graph of the Faculty of Architecture and the Built Environment, location tracking of different users is implemented and visualised in a heat-map. Additionally, a dashboard is created that can be used by a facility manager to translate the user paths to valuable information and reveal different movement patterns in an indoor facility.
The followed methodology showed promising results, concerning the reliability of ceilings for real-time indoor localisation, based on LiDAR and camera sensors, that are incorporated in up-to-date mobile devices. The robustness of Colored Iterative Closest Point (ICP) algorithm for indoor localisation based on point clouds was revealed, both in terms of time efficiency and quality, while the combination of Speeded-Up Robust Features (SURF) feature detector and Scale Invariant Feature Transform (SIFT) descriptor provides the optimal indoor localisation results with image data. The proposed pipeline revealed encouraging results for use in emergency situations, based on static data acquisition of a user, while it is also suitable for dynamic applications, in case a sensor is mounted on an automated device for indoor mapping operations.
Therefore, this thesis investigates the possibility of the ceilings in public or semi-public buildings, being used for indoor localisation, by using features that are included in a simple mobile device. The research additionally involves location tracking of different users, in order to discover different movement patterns in an indoor facility. Indoor localisation is achieved based on the comparison of user and reference data, that can be both point clouds and images, using the Light detection and ranging (LiDAR) of an iPad 12 pro and camera sensors of an Android device. The point cloud-based localisation is implemented based on different combinations of global and local registration techniques, while the image-based approach involves different feature detection, description and matching techniques. Using a web-application to visualise the indoor localisation results, an indoor model and a network graph of the Faculty of Architecture and the Built Environment, location tracking of different users is implemented and visualised in a heat-map. Additionally, a dashboard is created that can be used by a facility manager to translate the user paths to valuable information and reveal different movement patterns in an indoor facility.
The followed methodology showed promising results, concerning the reliability of ceilings for real-time indoor localisation, based on LiDAR and camera sensors, that are incorporated in up-to-date mobile devices. The robustness of Colored Iterative Closest Point (ICP) algorithm for indoor localisation based on point clouds was revealed, both in terms of time efficiency and quality, while the combination of Speeded-Up Robust Features (SURF) feature detector and Scale Invariant Feature Transform (SIFT) descriptor provides the optimal indoor localisation results with image data. The proposed pipeline revealed encouraging results for use in emergency situations, based on static data acquisition of a user, while it is also suitable for dynamic applications, in case a sensor is mounted on an automated device for indoor mapping operations.