Creating 3D city models with building footprints and LIDAR point cloud classification: A machine learning approach

The increasing availability of open geospatial data, such as building footprint vector data and LiDAR (Light Detection and Ranging) point clouds, has provided opportunities to generate large-scale 3D city models at low cost. However, using unclassified point clouds with building footprints to estimate building heights may yield erroneous results due to potential errors and anomalies in both datasets and their integration. Some of the points within footprints often reflect irrelevant objects other than roofs, leading to biases in height estimation, and few studies have developed systematic methods to filter them out. In this paper, a LiDAR point classification methodology is proposed that extracts only rooftop points for building height estimation. The LiDAR points are characterized by point, footprint, and neighborhood-based features and classified by the Random Forest (RF) algorithm into four classes – rooftop, wall, ground, and high outlier. The percentage of correctly classified points among 15,577 sample points in Columbus, Ohio, amounts to 96.5%. Conducting this classification separately for different building types (commercial, residential, skyscraper, and small constructions) does not significantly change the overall accuracy. The footprint-based features contribute most to predicting the classes correctly. Height validation results based on a sample of 498 buildings show that (1) using average or median heights with classified points provides the best estimates, minimizing the disparities between computed heights and ground truth and (2) the RF method yields outcomes much closer to ground truth than earlier classification approaches. Some outcomes are visualized in 3D format using Google Earth 3D Imagery and ArcScene.     

基于形态学建筑物指数的城市建筑物提取及其高度估算

通过引入高分辨率影像的形态学建筑物指数和阴影指数,并结合面向对象的地物信息提取思想,准确地提取出城市建筑物及其阴影,进而实现了城市建筑物的高度估算。首先,利用形态学建筑物指数的多方向多尺度特征,将建筑物与邻近光谱相似的道路目标进行分离;其次,采用双阈值策略提取建筑物与相应的阴影,进一步提高了建筑物的提取精度;最后,根据成像时刻卫星和太阳的高度角、方位角,建立建筑物阴影长度与建筑物高度的估算模型。试验利用厦门市思明区软件园资源三号(ZY\|3)数据进行城市建筑物提取及其高度估算,证实该方法能够较准确地估算出建筑物的高度信息,并且比基于SVM的监督分类方法具有更高的建筑物提取精度,建筑物高度估算的中误差可达±1 m。     

基于各向异性Markov随机场的遥感影像亚像元尺度建筑物提取

基于遥感影像的建筑物自动提取方法容易受混合像元影响,目标提取精度不高。亚像元定位可以提取亚像元尺度地物分布信息,减轻混合像元对目标提取结果造成的影响。传统亚像元定位模型采用各向同性邻域描述地物的空间相关性,并没有考虑地物特有的形状信息,难以满足建筑物提取的需要。在考虑建筑物光谱特征的基础上,建立了平行与垂直于目标建筑物主方向的各向异性邻域,并采用基于各向异性Markov随机场的亚像元定位模型进行了亚像元尺度的建筑物提取。基于QuickBird多光谱数据与AVIRIS高光谱数据的实验结果表明,该模型提取的建筑物不仅具有更高的空间分辨率,而且能够较好地保持建筑物边缘与角点的形状信息,是一种有效的亚像元尺度建筑物提取方法。     

Benchmarking stereo-derived DSMs to lidar by land cover and slope for resource development in northern Ontario,Canada

Topographic and elevation data are essential in the development of supporting infrastructure around mining sites. The de facto standard for acquiring elevation data is through light detection and ranging (lidar). The high labour and monetary cost of acquiring lidar has fostered more cost-effective approaches for creating elevation models that use stereo photogrammetry. To assess the accuracy of stereo-photogrammetry-derived elevation models and their potential application, we benchmark satellite (Worldview-2) and aircraft (South Central Ontario Orthoimagery Project; SCOOP) stereo-derived digital surface models (DSMs) against a lidar-derived DSM. Our results show that both stereo-derived DSMs have strong monotonic correlations with lidar across a range of land-cover types and slopes. The overall vertical accuracy of Worldview-2 and SCOOP DSMs are similar and do not meet the United States National Digital Elevation Program (NDEP) standards. However, accuracy assessment across land-cover types and slope categories show that specific land cover types (i.e. grass, row crops/pasture, sparse vegetation and marsh) on gently sloping terrain compare well to lidar data and meet NDEP accuracy standards. We situate the presented research in the context of northern resource development and discuss opportunities to improve the vertical accuracy of stereo-derived DSMs, for example, through unmanned aerial systems.     

Urban building roof segmentation from airborne lidar point clouds

This article presents a new approach to segmenting building rooftops from airborne lidar point clouds. A progressive morphological filter technique is first applied for separation between ground and non-ground points. For the non-ground points, a region-growing algorithm based on a plane-fitting technique is used to separate building points from vegetation points. Then, an adaptive Random Sample Consensus (RANSAC) algorithm based on a grid structure is developed to improve the probability of selecting an uncontained sample from the localized sampling. The distance, standard deviation and normal vector are integrated to keep topological consistency among building rooftop patches during building rooftop segmentation. Finally, the remaining points are mapped on to the extracted planes by a post-processing technique to improve the segmentation accuracy. The results for buildings with different roof complexities are presented and evaluated.     

GEDI地面高程和森林冠层高度的精度评价与影响分析

精确地提取地面高程和植被冠层高度,对于地形地貌、生态学等方面的研究具有重要意义。2018年12月发射的新一代全球生态系统动力学调查雷达（GEDI）为地面高程和植被冠层高度大范围精确提取提供了前所未有的机会。研究旨在利用机载激光雷达数据验证GEDI提取的地面高程和冠层高度精度,并探讨地理定位误差、地形坡度、坡向、植被覆盖度、方位角、采集时间、光束类型和不同森林类型因素对其精度的影响。结果表明：通过校正GEDI数据地理定位误差,可以明显提高其提取的地面高程和冠层高度精度;影响冠层高度提取精度最主要的因素是植被覆盖度,其次是坡度;影响地面高程提取精度的主要因素为坡向、坡度。植被覆盖度大于25%时,数据精度更高;坡度为0°—5°的缓坡地区地面高程和冠层高度精度最高。该研究结果将为GEDI数据筛选与应用提供依据。     

Trend analysis of building height and total floor space in Beijing,China using ICESat/GLAS data

GLA14, one of the products of the spaceborne Light Detection and Ranging (LiDAR) sensor Geoscience Laser Altimeter System (GLAS), provides six Gaussian decomposition waveforms that represent different vertical layers of the ground target in a laser spot. In this article, we have extracted the relative height of ground targets from peak positions of the GLAS waveform, carried out the field validations, analysed the trend of building height in Beijing and then multiplied the building height and the percentage of building area within a pixel of the land-use/land-cover classification map to get the annual change of total floor space of buildings in Beijing. Based on the total floor space of buildings (TFSB) released by the National Bureau of Statistics of China (NBSC), we have established a linear regression model between the GLAS-estimated total floor space in Beijing and the data provided by NBSC. The results show that the building height and (TFSB) in Beijing increased from 2003 to 2008. The method proposed in this article expands research on urban change from a two-dimensional plane to a three-dimensional space to improve research accuracy, and is complementary to current remote-sensing methods.     

Synoptic mapping of high-rise buildings in urban areas based on combined shadow analysis and scale space processing

A highly automated methodology is described to map locations and heights of high-rise buildings from single high-resolution multi-spectral satellite imagery. The approach involves preliminary shadow detection using the Tsai colour invariant transform and scale space processing to identify candidate building pixels. Application of shadow-building and shadow length constraints led to mapping of the location and height of building candidate objects. The approach has been applied to a winter SPOT 5 scene of Beijing, China. Tests of buildings in a suburban area indicate that a high detection rate (93%) can be achieved for buildings taller than 28 m. A height estimation accuracy of 20 m has also been met for these buildings.     

An Automation System of Rooftop Detection and 3D Building Modeling from Aerial Images

This paper presents a prototype system of rooftop detection and 3D building modeling from aerial images. In this system, without the knowledge of the position and orientation information of the aerial vehicle a priori, the parameters of the camera pose and ground plane are first estimated by simple human?Ccomputer interaction. Next, after an over-segmentation of the aerial image by the Mean-Shift algorithm, the rooftop regions are coarsely detected by integrating multi-scale SIFT-like feature vectors with SVM-based visual object recognition. 2D cues alone however might not always be sufficient to separate regions such as parking lots from building roofs. Thus in order to further refine the accuracy of the roof-detection result and remove the misclassified non-rooftop regions such as parking lots, we further resort to 3D depth information estimated based on multi-view geometry. More specifically, we determine whether a candidate region is a rooftop or not according to its height information relative to the ground plane, whereas the candidate region??s height information is obtained by a novel, hierarchical, asymmetry correlation-based corner matching scheme. The output of the system will be a water-tight triangle mesh based 3D building model texture mapped with the aerial images. We developed an interactive 3D viewer based on OpenGL and C+?+ to allow the user to virtually navigate the reconstructed 3D scene with mouse and keyboard. Experimental results are shown on real aerial scenes.     

基于建筑物提取的精细尺度人口估算研究

以精细尺度的人口估算为目标,提出一种根据居民区建筑物属性估算人口数量的方法。首先基于Dempster-Shafer证据理论,结合LiDAR数据和高分辨率遥感影像进行建筑物的自动提取。根据土地利用分类图排除提取结果中的非居民区建筑后,按照线性回归的思想,通过对居民建筑物的数量、面积、体积等几何属性的优化选择建立人口估算模型。实验表明,利用该估算模型能够获得较高精度的小面积目标区域上的估算结果。该方法提高了人口估算的精细程度和自动化程度。     

Investment estimation of prefabricated concrete buildings based on XGBoost machine learning algorithm

The prefabricated concrete buildings (PCBs）are the booster in the process of construction industrialization and intelligent upgrading. However, its high cost has become one of the restricting factors of further application and promotion of prefabricated concrete buildings. Moreover, the existing investment estimation methods of prefabricated concrete buildings have limited predicting accuracy as well as the ability of adapting dynamic factors. Therefore, to achieve more reliable and reasonable investment estimation of prefabricated concrete buildings, this paper has proposed an investment estimation model of prefabricated concrete buildings based on XGBoost machine learning algorithm. In the proposed model, the construction project cost-significance theory (CS) and analytic hierarchy process (AHP) were used to extract the construction characteristic indices of prefabricated concrete buildings investment estimation. Then the XGBoost machine learning algorithm was implemented to build an investment estimation model of prefabricated concrete buildings that was able to quantify the uncertainty of the confidence and prediction, and to enhance the interpretability of the model. The research conducted in this paper showed that when compared with traditional machine learning methods such as Support vector machine (SVM), Back Propagation Neural Network (BPNN) and Random Forest (RF), XGBoost had better generalization and interpretable ability. The discussion provided in this paper further demonstrated the reliability and feasibility of the proposed model, and provided reliable basis for the investment decision-making of prefabricated concrete building projects.     

Detection and typification of linear structures for dynamic visualization of 3D city models

Cluttering is a fundamental problem in 3D city model visualization. In this paper, a novel method for removing cluttering by typification of linear building groups is proposed. This method works in static as well as dynamic visualization of 3D city models. The method starts by converting building models in higher Levels of Details (LoDs) into LoD1 with ground plan and height. Then the Minimum Spanning Tree (MST) is generated according to the distance between the building ground plans. Based on the MST, linear building groups are detected for typification. The typification level of a building group is determined by its distance to the viewpoint as well as its viewing angle. Next, the selected buildings are removed and the remaining ones are adjusted in each group separately. To preserve the building features and their spatial distribution, Attributed Relational Graph (ARG) and Nested Earth Mover’s Distance (NEMD) are used to evaluate the difference between the original building objects and the generalized ones. The experimental results indicate that our method can reduce the number of buildings while preserving the visual similarity of the urban areas.     

Effects of Different Spatial Resolution of Remote SensingImages on Estimation Accuracy of Urban Building Height

In this study,the remote sensing images of WorldView-2,GF-2,and GF-1,which cover Xiamen Software Park,were selected for study.A building and shadow extraction process suitable for different images was constructed,which applied object\|oriented approach and morphology ideas combined with spectral,shadow and shape constraints.Subsequently,the building heights of three different spatial resolutions of 0.5 m,1 m and 2 m were estimated by using the shadow length estimation method.Finally,the influence of image spatial resolution on building extraction accuracy and building height estimation accuracy was evaluated quantitatively.The main conclusions are as follows：(1) The improved building and shadow extraction process achieves higher extraction accuracy,but accuracy decreases slightly with the decrease of spatial resolution of images;(2) With the decrease of spatial resolution,the accuracy of building height estimation decreases gradually,but it does not show linear relationship.At the resolution increases from 1m to 0.5 m,the accuracy of building height estimation increases faster than the resolution increases from 2 m to 1 m;(3) GF-1 is more suitable for height estimation of high\|rise buildings and GF-2 is suitable for middle and high rise buildings,while WorldView\|2 has higher estimation accuracy for building height in different height ranges.     

Building detection in an urban area using lidar data and QuickBird imagery

This article presents a hierarchical approach to detect buildings in an urban area through the combined usage of lidar data and QuickBird imagery. A normalized digital surface model (nDSM) was first generated on the basis of the difference between a digital surface model and the corresponding digital terrain model. Then, ground objects were removed according to a height threshold. In consideration of the relief displacement effect in very high resolution remote-sensing imagery, we segmented the nDSM by the region-growing method and used the overlap ratio to avoid over-removing building objects. Finally, the region size and spatial relation of trees and buildings were used to filter out trees occluded by buildings based on an object-based classification. Compared with previous methods directly using the normalized difference vegetation index (NDVI), our method improved the completeness from 85.94% to 90.20%. The overall accuracy of the buildings detected using the proposed method can be up to 94.31%, indicating the practical applicability of the method.     

Virtual first and last pulse method for building detection from dense LiDAR point clouds

ABSTRACT

LiDAR (light detection and ranging) technology is one of the most important techniques used in photogrammetry and remote sensing in order to extract high quality and high density 3D point clouds. In this paper, a novel method is proposed to detect buildings without filtering the ground points, depending on a new method named Virtual First and Last Pulse (VFLP). The type of pulses is reclassified according to the vertical direction, so that the first and last pulses for each imaginary vertical column are extracted. Using the height difference between the virtual first pulse (VFP) and virtual last pulse (VLP), the vertical features can be extracted. One of these features is the building walls, which are used as a mask for the building and, in turn, is used to detect the buildings. The results show that this method is very effective for detecting buildings and removing the trees and vegetation without filtering the ground points. Also, this method obtained a promising result for per-area and per-object level. The results show the completeness of 98.75%, correctness of 97.29% and quality of 96.1% at the per-area level and completeness of 91.95%, correctness of 98.63% and quality of 90.79% at the per- object level.     

结合体元数据结构的机载LIDAR建筑物检测

目的 目前,点云、栅格格网及不规则三角网等建筑物检测中常用的离散机载激光雷达（LIDAR）点云数据表达方式存在模型表达复杂、算法开发困难、结果表达不准确及难以表达多返回数据等缺点。为此,针对LIDAR点云体元结构模型构建及在此基础上的建筑物检测展开研究,提出一种基于体元的建筑物检测算法。方法 首先将点云数据规则化为二值（即1、0值,分别表示体元中是否包含有激光点）3D体元结构。然后利用3D滤波算法将上述体元结构中表征数据点的体元分类为地面和非地面体元。最后,依据建筑物边缘的接近直线、跳变特性从非地面体元中搜寻建筑物边缘作为种子体元进而标记与其3D连通的非地面体元集合为建筑物体元。结果 实验基于ISPRS（international society for photogrammetry and remote sensing）提供的包含了不同的建筑物类型的城区LIDAR点云数据测试了"邻域尺度"参数的敏感性及提出算法的精度。定量评价的结果表明：56邻域为最佳邻域尺度;建筑物的检测质量可达到95%以上——平均完整度可达到95.61%、平均正确率可达95.97%。定性评价的结果表明：对大型、密集、不规则形状、高低混合及其他屋顶类型比较特殊的复杂建筑物均可成功检测。结论 本文提出的建筑物检测算法采用基于体元空间邻域关系的搜索标记方式,可有效实现对各类建筑目标特别是城市建筑目标的检测,检测结果易于建模3D建筑物模型。     

Large-scale parameterization of 3D building morphology in complex urban landscapes using aerial LiDAR and city administrative data

The form and function of the modern city are defined by the three-dimensional contours of the built environment. The morphology of the urban landscape has significant implications for a city's sustainability, efficiency, and resilience. With advancements in remote sensing, especially airborne Light Detection and Ranging (LiDAR), the potential exists to model urban topography at an unprecedented spatial resolution and granularity and extract previously unavailable characteristics of individual buildings. In this study, we demonstrate the application of point-based voxelization techniques to extract design parameters in complex urban environments at unprecedented scale using New York City, and its more than 1,000,000 buildings, as a test case. Covering approximately 800 km², we develop a 1 m² resolution Digital Surface Model (DSM) derived from aerial LiDAR point cloud data, together with city administrative records, to calculate building massing, height, volume, exposed surface area, and compactness ratios for every building in the City. The proposed scalable approach creates a significant opportunity for city administrators, urban planners, architectural engineers, and building designers to understand the relationship between urban morphology and a range of infrastructure and environmental systems.     

抗野值鲁棒滤波在微惯性组合导航中的应用

在鲁棒H∞滤波应用过程中,如果量测序列含有野值,将会严重影响滤波精度。针对这一问题提出一种新的剔除野植的方法;从新息入手,首先利用小波变换系数特性,通过最细尺度上的小波系数来检测野值点,然后基于信息扩散原理,采用替代方法,对含有单个或连续野值的新息加以修正,从而达到检测和剔除野值的目的。通过对基于MEMS的车载微惯性SINS/GPS组合导航的仿真表明,新算法能够有效的检测出野值,并在野值单个或成片出现的情况下都能保证滤波精度。     

基于迁移学习的低空摄影测量滑坡方量估算方法

针对现有方法难以准确地估算山体滑坡体积的问题,引入人工智能算法,提出耦合迁移学习与微分算法的低空摄影测量山体滑坡方量估算方法。首先,利用SfM与SGM密集匹配等算法从低空无人机立体影像中解算出高精度三维密集点云,结合可见光植被指数和双边滤波算法从密集点云中剥离出目标区地面点云;然后,构建深度神经网络插值模型来表征二维坐标与高程之间的非线性映射关系,并基于参数共享的迁移学习来自适应优化深度神经网络以实现滑坡目标区高程值预测,进而重构滑坡区域的数字地表模型;最后,基于目标区滑坡前后数字地表模型高程差值和微分算法实现山体滑坡方量估算。实验结果表明,该方法平均相对误差为2.7%,相比常用的方法,显著提高了滑坡方量估计精度,并能适应不同地形条件下滑坡方量估算。     

多特征融合的高分辨率遥感影像建筑物分级提取

目的 针对高分辨率遥感影像普遍存在的同谱异物和同物异谱问题,提出一种综合利用光谱、形状、空间上下文和纹理特征的建筑物分级提取方法。方法 该方法基于单幅高分辨率遥感影像,首先利用多尺度多方向梯度算子构造的建筑物指数和形状特征提取部分分割完整的矩形建筑物目标;然后由多方向线性结构元素和形态学膨胀运算确定投票矩阵,从而获取光照方向,并利用光照方向和阴影特征对已提取建筑物进行筛选,剔除非建筑物对象,完成建筑物初提取;最后借助初提取建筑物对象的纹理特征向量建立概率模型,取得像素级建筑物提取结果,将该结果与影像分割相结合实现建筑物提取。结果 选取两幅高分辨率遥感影像进行实验,在建筑物初提取实验中,将本文方法与邻域总变分法和Sobel算子进行对比,实验结果表明,本文方法适用性强,为后提取提供的建筑物样本可靠性更高。在建筑物提取实验中,采用查准率、查全率和F₁分数3个指标进行定量分析,与形态学建筑物指数结合形态学阴影指数算法、邻域总变分结合混合高斯模型和贝叶斯判决算法相比,各项精度指标均得到显著提升,其中查准率提高了2.90个百分点,查全率提高了12.49个百分点,F₁分数则提升了8.84。结论 本文提出的建筑物分级提取方法具备一定抗干扰能力,且提取准确性高,适用性强。