The integration of aerial and ground images is known to be effective for enhancing the quality of 3D reconstruction in complex urban scenarios. However, directly applying the structure-from-motion (SfM) technique for unified 3D reconstruction with aerial and ground images is particularly difficult, due to the large differences in viewpoint, scale, and appearance between those two types of images. Previous studies mainly rely on viewpoint rectification or view rendering/synthesis to improve the feature matching quality for aligning the aerial and ground models. Nevertheless, these approaches still fail to address the inherent information differences between aerial and ground images. In this paper, we propose a learning-based matching framework for direct SfM with ground and aerial images. The key idea of our method is to learn the pixel-wise consistent features between aerial and ground images to handle the large heterogeneity of these two types of images. Specifically, we deploy a learning-based matching framework to robustly correspond the aerial and ground images. With the high-quality feature matching, learned feature maps are used for refining keypoint locations and fusing featuremetric error into bundle adjustment with the consideration of geometric error, both of which can further improve the accuracy and completeness of recovered 3D scene. Extensive experiments conducted on six datasets demonstrate that the proposed method can reconstruct high-fidelity 3D models with direct aerial-to-ground SfM, which cannot be achieved by existing methods. In addition, our method also shows outstanding performance in sub-tasks of feature matching and point cloud recovery.

Curvilinear building forms are increasingly common in modern architecture, yet their impact on wind dynamics remains understudied. This research examines the wind flow behavior around high-rise residential building rows configured with convex and concave curvilinear shapes, focusing on the influence of varying central angles. Using parametric models generated in Rhino’s Grasshopper, followed by experimentally validated CFD simulations, we analyze the aerodynamic effects of these configurations. The findings indicate contrasting impacts: convex curvilinear rows amplify wind flow as their central angles increase, while concave rows hinder airflow under similar conditions. These variations are linked to changes in the windward projective areas. This study enhances the understanding of wind behavior in urban settings and provides key insights for optimizing natural ventilation and designing sustainable curvilinear building clusters.

Although T-form buildings have been widely observed in newly constructed high-rise residential communities, there have been relatively limited investigations into the influence of their geometries on wind environments. This study aims to address this gap by conducting iterative quantitative assessments of the influences of various bulge-part sizes of typical T-form high-rise residential buildings on surrounding wind environments. A methodology has been employed by integrating multiple computational tools, including parametric modeling, Computational Fluid Dynamics (CFD), and Influenced Areas (IAs) analysis. Representative T-form buildings have been modeled with parametric components, allowing for easy variation of bulge-part sizes. The investigation process involves sequential steps of parametric modeling, experimentally validated CFD simulations, statistical assessment, and subsequent results analysis and discussions. Findings could be summarized as follows: (1) according to IAs analysis, the influences on wind environments were decreased as the bulge-part sizes were increased, and the decrease of the bulge-part sizes could cause the contrary effect; (2) the promotion of outdoor ventilation caused by the increase of the bulge-part length was more than the increase of the bulge-part width according to the correlation coefficients (0.88 > 0.78; 0.88 > 0.76); (3) it was recommended to design relatively large bulge parts on the windward side to enhance outdoor ventilation. The research outcomes provide valuable and insightful information for the development of sustainable architectural design strategies aimed at optimizing natural ventilation.

Wind environment is important in architectural sustainable design, as existing studies show that it can be considerably influenced by building morphologies. This study aimed to develop a data-mining framework to quantitatively evaluate and compare influences on Low-Wind-Velocity Area (LWVA) of common cuboid-form buildings with typical morphological parameters. The data-mining framework was originally developed by integrating multiple computational methods for rapid in-depth iterative analyses, including the generation of building models using parametric modelling, the big data generation based on hybrid model, and the statistical metric analysis method. The hybrid model was created by combining the CFD model and machine learning model. The accuracy and efficiency of the framework were fully demonstrated through the comprehensive validation and analyses of different models. The data of more than fifty thousand building cases with different morphological parameters and relevant wind conditions were generated and analyzed. Influences on LWVA of morphological parameters of cuboid-form building was comprehensively evaluated, including the visualization of multiple parameters, calculation and comparison of several correlation coefficients. It suggested that the reduction of height and width on the windward side would significantly decrease the LWVA and promote the outdoor ventilation. The change of depth would have relatively limited influence on the LWVA. Multivariate regression model-fit and variance analyses were further implemented, and it was found that there was a relatively significant linear correlation between the LWVA and morphological parameters. The equation of multivariate regression model was provided for extra rapid prediction. The study outcome could contribute to efficient evaluation of LWVA and provide useful information for sustainable design.

3D (three-dimensional) property volume is an important data carrier for 3D land administration by using 3D cadastral technology, which can be used to express the legal space (property rights) scope matching with physical entities such as buildings and land. A 3D property volume is represented by a dense set of 3D coordinate points arranged in a predefined order and is displayed alongside the parcel map for reference and utilization by readers. To store a 3D property volume in the database, it is essential to record the connectivity relationships among the original 3D coordinate points, the associations between points and lines for representing boundary lines, and the relationships between lines for defining surfaces. Only by preserving the data structure that represents the relationships among points, lines, and surfaces can the 3D property volume in a parcel map be fully reconstructed. This approach inevitably results in the database storage volume significantly exceeding the original size of the point set, thereby causing storage redundancy. Consequently, this paper introduces a reversible 3D property volume compression coding method (called 3DPV-CC) to address this issue. By analyzing the distribution characteristics of the coordinate points of the 3D property volume, a specific rule for sorting the coordinate points is designed, enabling the database to have the ability of data storage and recovery by merely storing a reordered point set. The experimental results show that the 3DPV-CC method has excellent support capabilities for 3D property volumes of the vertical and slopped types, and can compress and restore the coordinate point set of the 3D property volume for drawing 3D parcel maps. The compression capacity of our method in the test is between 23.66% and 38.42%, higher than the general data compression methods (ZIP/7Z/RAR: 8.37–10.32%). By means of this method, land or real estate administrators from government departments can store 3D property volume data at a lower cost. This is conducive to enhancing the informatization level of land management.

Urban building information reflects the status and trends of a region’s development and is essential for urban sustainability. Detection of buildings from high-resolution (HR) remote sensing images (RSIs) provides a practical approach for quickly acquiring building information. Mainstream building detection methods are based on fully supervised deep learning networks, which require a large number of labeled RSIs. In practice, manually labeling building instances in RSIs is labor-intensive and time-consuming. This study introduces semi-supervised deep learning techniques for building detection and proposes a semi-supervised building detection framework to alleviate this problem. Specifically, the framework is based on teacher–student mutual learning and consists of two key modules: the color and Gaussian augmentation (CGA) module and the consistency learning (CL) module. The CGA module is designed to enrich the diversity of building features and the quantity of labeled images for better training of an object detector. The CL module derives a novel consistency loss by imposing consistency of predictions from augmented unlabeled images to enhance the detection ability on the unlabeled RSIs. The experimental results on three challenging datasets show that the proposed framework outperforms state-of-the-art building detection methods and semi-supervised object detection methods. This study develops a new approach for optimizing the building detection task and a methodological reference for the various object detection tasks on RSIs.