关键点检测的复杂建筑物模型自动重建

利用机载LiDAR点云数据进行建筑物重建是当今摄影测量与遥感领域的一个热点问题,特别是复杂形状建筑物模型的精确自动构建一直是一个难题。本文提出一种基于关键点检测的复杂建筑物模型自动重建方法,采用RANSAC法与距离法相结合的分割方法自动提取建筑物屋顶各个平面的点云,并利用Alpha Shape算法提取出各个平面的精确轮廓,根据屋顶平面之间的空间拓扑关系分析建筑物的公共交线特征,在此特征约束下对提取的初始关键点进行修正,最终重建出精确的建筑物3维模型。选取不同类型复杂建筑物与包含复杂建筑物的城市区域点云进行实验,结果表明该算法具有较强实用价值。     

Automatic extraction of building roofs using LIDAR data and multispectral imagery

Automatic 3D extraction of building roofs from remotely sensed data is important for many applications including city modelling. This paper proposes a new method for automatic 3D roof extraction through an effective integration of LIDAR (Light Detection And Ranging) data and multispectral orthoimagery. Using the ground height from a DEM (Digital Elevation Model), the raw LIDAR points are separated into two groups. The first group contains the ground points that are exploited to constitute a ‘ground mask’. The second group contains the non-ground points which are segmented using an innovative image line guided segmentation technique to extract the roof planes. The image lines are extracted from the grey-scale version of the orthoimage and then classified into several classes such as ‘ground’, ‘tree’, ‘roof edge’ and ‘roof ridge’ using the ground mask and colour and texture information from the orthoimagery. During segmentation of the non-ground LIDAR points, the lines from the latter two classes are used as baselines to locate the nearby LIDAR points of the neighbouring planes. For each plane a robust seed region is thereby defined using the nearby non-ground LIDAR points of a baseline and this region is iteratively grown to extract the complete roof plane. Finally, a newly proposed rule-based procedure is applied to remove planes constructed on trees. Experimental results show that the proposed method can successfully remove vegetation and so offers high extraction rates.     

Automatic filtering and 2D modeling of airborne laser scanning building point cloud

This article suggests a new approach to automatic building footprint modeling using exclusively airborne LiDAR data. The first part of the suggested approach is the filtering of the building point cloud using the bias of the Z‐coordinate histogram. This operation aims to detect the points of roof class from the building point cloud. Hence, eight rules for histogram interpretation are suggested. The second part of the suggested approach is the roof modeling algorithm. It starts by detecting the roof planes and calculating their adjacency matrix. Hence, the roof plane boundaries are classified into four categories: (1) outer boundary; (2) inner plane boundaries; (3) roof detail boundaries; and (4) boundaries related to the missing planes. Finally, the junction relationships of roof plane boundaries are analyzed for detecting the roof vertices. With regard to the resulting accuracy quantification, the average values of the correctness and the completeness indices are employed in both approaches. In the filtering algorithm, their values are respectively equal to 97.5 and 98.6%, whereas they are equal to 94.0 and 94.0% in the modeling approach. These results reflect the high efficacy of the suggested approach.     

Segmentation of Airborne Point Cloud Data for Automatic Building Roof Extraction

Roof plane segmentation is a complex task since point cloud data carry no connection information and do not provide any semantic characteristics of the underlying scanned surfaces. Point cloud density, complex roof profiles, and occlusion add another layer of complexity which often encounter in practice. In this article, we present a new technique that provides a better interpolation of roof regions where multiple surfaces intersect creating non-manifold points. As a result, these geometric features are preserved to achieve automated identification and segmentation of the roof planes from unstructured laser data. The proposed technique has been tested using the International Society for Photogrammetry and Remote Sensing benchmark and three Australian datasets, which differ in terrain, point density, building sizes, and vegetation. The qualitative and quantitative results show the robustness of the methodology and indicate that the proposed technique can eliminate vegetation and extract buildings as well as their non-occluding parts from the complex scenes at a high success rate for building detection (between 83.9% and 100% per-object completeness) and roof plane extraction (between 73.9% and 96% per-object completeness). The proposed method works more robustly than some existing methods in the presence of occlusion and low point sampling as indicated by the correctness of above 95% for all the datasets.     

Building roof modeling from airborne laser scanning data based on level set approach

This paper presents a novel approach to building roof modeling, including roof plane segmentation and roof model reconstruction, from airborne laser scanning data. Segmentation is performed by minimizing an energy function formulated as multiphase level set. The energy function is minimized when each segment corresponds to one or several roof plans of the same normal vector. With this formulation, maximum n regions are segmented at a time by applying log₂n level set functions. The roof ridges or step edges are then delineated by the union of the zero level contours of the level set functions. In the final step of segmentation, coplanar and parallel roof segments are separated into individual roof segments based on their connectivity and homogeneity. To reconstruct a 3D roof model, roof structure points are determined by intersecting adjacent roof segments or line segments of building boundary and then connected based on their topological relations inferred from the segmentation result. As a global solution to the segmentation problem, the proposed approach determines multiple roof segments at the same time, which leads to topological consistency among the segment boundaries. The paper describes the principle and solution of the multiphase level set approach and demonstrates its performance and properties with two airborne laser scanning data sets.     

Automated Detection of 3D Roof Planes from Lidar Data

The purpose of this study is to derive vectoral 3D roof planes from the LIDAR point cloud of the detected buildings. For segmentation of the LIDAR point cloud, the RANSAC algorithm has been used. Because the RANSAC algorithm is sensitive to the used parameters, and results in over- or under-segmentation of the clusters, a refinement method has been proposed. The detection of roof planes has been improved with use of the refinement method. Therefore, similar plane surfaces have been combined, followed by the region-growing algorithm, to split the under-segmented plane surfaces. The digitization of the roof boundaries is performed using the alpha-shapes algorithm, followed by line fitting to generalize the roof edges. The quality assessment has been done using the reference vector dataset with comparison using four different criteria.     

Robust statistical approaches for local planar surface fitting in 3D laser scanning data

This paper proposes robust methods for local planar surface fitting in 3D laser scanning data. Searching through the literature revealed that many authors frequently used Least Squares (LS) and Principal Component Analysis (PCA) for point cloud processing without any treatment of outliers. It is known that LS and PCA are sensitive to outliers and can give inconsistent and misleading estimates. RANdom SAmple Consensus (RANSAC) is one of the most well-known robust methods used for model fitting when noise and/or outliers are present. We concentrate on the recently introduced Deterministic Minimum Covariance Determinant estimator and robust PCA, and propose two variants of statistically robust algorithms for fitting planar surfaces to 3D laser scanning point cloud data. The performance of the proposed robust methods is demonstrated by qualitative and quantitative analysis through several synthetic and mobile laser scanning 3D data sets for different applications. Using simulated data, and comparisons with LS, PCA, RANSAC, variants of RANSAC and other robust statistical methods, we demonstrate that the new algorithms are significantly more efficient, faster, and produce more accurate fits and robust local statistics (e.g. surface normals), necessary for many point cloud processing tasks. Consider one example data set used consisting of 100 points with 20% outliers representing a plane. The proposed methods called DetRD-PCA and DetRPCA, produce bias angles (angle between the fitted planes with and without outliers) of 0.20° and 0.24° respectively, whereas LS, PCA and RANSAC produce worse bias angles of 52.49°, 39.55° and 0.79° respectively. In terms of speed, DetRD-PCA takes 0.033 s on average for fitting a plane, which is approximately 6.5, 25.4 and 25.8 times faster than RANSAC, and two other robust statistical methods, respectively. The estimated robust surface normals and curvatures from the new methods have been used for plane fitting, sharp feature preservation and segmentation in 3D point clouds obtained from laser scanners. The results are significantly better and more efficiently computed than those obtained by existing methods.     

激光扫描中平面拟合及坐标转换模型构建

坐标转换是测绘领域经常遇到的问题,如局部坐标系和全局坐标系间的转换、多传感器融合时各传感器间坐标系的相互转换等等.采用平面代替传统坐标转换中的公共点,建立了基于平面的坐标转换模型.首先给出平面的标准定义;然后提出一种基于稳健的 RANSAC算法的特征值的平面拟合方法用于从点云数据中提取公共平面;构建基于平面表示的坐标转换模型并推导转换参数的计算方法;最后,通过模拟数据与实测数据验证模型的正确性.     

Tree Detection in Urban Regions Using Aerial Lidar and Image Data

In this letter, we present an approach to detecting trees in registered aerial image and range data obtained via lidar. The motivation for this problem comes from automated 3-D city modeling, in which such data are used to generate the models. Representing the trees in these models is problematic because the data are usually too sparsely sampled in tree regions to create an accurate 3-D model of the trees. Furthermore, including the tree data points interferes with the polygonization step of the building roof top models. Therefore, it is advantageous to detect and remove points that represent trees in both lidar and aerial imagery. In this letter, we propose a two-step method for tree detection consisting of segmentation followed by classification. The segmentation is done using a simple region-growing algorithm using weighted features from aerial image and lidar, such as height, texture map, height variation, and normal vector estimates. The weights for the features are determined using a learning method on random walks. The classification is done using the weighted support vector machines, allowing us to control the misclassification rate. The overall problem is formulated as a binary detection problem, and the results presented as receiver operating characteristic curves are shown to validate our approach     

利用层间连接和平滑策略重建屋顶三维模型

面向数字城市和智慧城市建设急需城市建筑三维模型支撑的需要,本文基于机载LiDAR数据,以“顾及平整性的屋顶面片分割—屋顶层间连接—三维模型重建”为脉络,提出了一种采用层间连接和平滑策略的建筑屋顶三维模型重建方法。在屋顶面片提取过程中,充分顾及了屋顶面片的平整性;并在屋顶面片平整基础上,提出层间连接点的概念,以实现高效、快速的模型重建工作。试验部分,本文从屋顶面片重建完整率与正确率、重建几何精度及建筑物高程对于重建的影响3个方面作了较为详尽的评价与分析,并在国际摄影测量与遥感学会标准数据集支撑下,与国际同行进行试验对比。试验结果表明,建筑屋顶重建的完整率和正确率分别达到90%和95%;在偏移距离评价方面,平均偏移距离和标准差最优分别达0.05 m和0.18 m。因此,本文方法可有效完成建筑屋顶三维模型重建,重建模型准确度高、完整性好。     

车载联合机载LiDAR点云数据的建筑物立面精细分割

从数据量庞大且散乱的车载LiDAR点云中分割出建筑物立面数据是一项繁琐而艰巨的工作。本文提出一种结合机载LiDAR点云的车载LiDAR点云建筑物立面分割方法。该方法在空-地点云严格配准的基础上,从机载LiDAR点云中分割出每栋建筑物的顶部点云,提取建筑物顶部外轮廓线并进行规则矢量化处理,设置轮廓线缓冲区实现立面点云的粗分割;再采用基于稳健特征值的平面拟合法对单栋建筑物的每个立面进行去噪滤波,实现建筑物立面的精细分割。试验结果证明了该算法对城市场景中车载LiDAR点云处理的有效性。     

Cycle graph analysis for 3D roof structure modelling: Concepts and performance

The paper presents a cycle graph analysis approach to the automatic reconstruction of 3D roof models from airborne laser scanner data. The nature of convergences of topological relations of plane adjacencies, allowing for the reconstruction of roof corner geometries with preserved topology, can be derived from cycles in roof topology graphs. The topology between roof adjacencies is defined in terms of ridge-lines and step-edges. In the proposed method, the input point cloud is first segmented and roof topology is derived while extracting roof planes from identified non-terrain segments. Orientation and placement regularities are applied on weakly defined edges using a piecewise regularization approach prior to the reconstruction, which assists in preserving symmetries in building geometry. Roof corners are geometrically modelled using the shortest closed cycles and the outermost cycle derived from roof topology graph in which external target graphs are no longer required. Based on test results, we show that the proposed approach can handle complexities with nearly 90% of the detected roof faces reconstructed correctly. The approach allows complex height jumps and various types of building roofs to be firmly reconstructed without prior knowledge of primitive building types.     

基于带权点法向量的LiDAR数据屋顶检测方法

提出了一种基于带权点法向量的LiDAR数据屋顶检测方法。通过利用点和其邻接点构成的面法向量进行峰值统计,检测屋顶面。检测过程中同时考虑每个点法向量的权值,从而确定每个点对面的贡献,一定程度上消除了噪声的影响,提高了小面积屋顶检测的准确程度。同时,采用多阈值进行屋顶面检测,能够检测大小不同的面。通过实验验证了本算法的有效性。     

利用多尺度张量投票的建筑立面分割方法

地面三维激光扫描是城市建筑立面数据采集的新方法,由于三维点云具有数据量大、无规则等特点,导致从点云中精确分割城市建筑物信息面临着严峻的挑战.在传统张量投票方法基础上,充分考虑各个尺度下点云立面特征判别的概率,提出了一种多尺度张量投票方法来实现平面分割,更加精确地实现了立面特征的识别.通过实例,将该方法同主成分分析和传统...     

基于局部约束的建筑物屋顶点云平面分割方法

建筑物屋顶面点云分割结果的好坏对建筑物三维模型重建起着重要的作用。针对传统RANSAC算法建筑物屋顶面点云的分割问题,提出了一种基于局部约束的建筑物点云平面分割方法。利用点云局部曲面法向约束构建法向准则,利用半径约束的点云空间聚类的方法对共面屋顶面点云进行分解,从而抑制"伪屋顶面"的产生;利用局部抽样策略降低算法的迭代次数,减少运算量。实验表明该方法能够获得稳定可靠的建筑物屋顶点云分割结果,将有利于后续的建筑物三维模型重建。     

基于分裂合并的多模型拟合方法在点云分割中的应用

本文基于机器视觉探讨数字摄影测量三维构像下的智能数据处理要素之二：海量点云分割处理技术。多模型拟合方法通过将点云拟合到不同模型中,依照点云空间分布特征和几何结构特征进行分割。针对点云数据量巨大、分布不均匀、结构复杂等特性,本文提出一种基于多模型拟合的点云分割方法。首先通过降采样,采用基于密度分布的聚类方法,实现对点云的预分割。在预分割基础上,利用基于分裂合并的多模型拟合方法对点云进行后续拟合分割。针对平面和弧面,本文采用不同的拟合方式,最终实现对室内密集点云分割。试验结果表明,该方法能够在无须提前设置模型数目的情况下实现点云的自动分割。且相较于现有的点云分割技术,此方法相较于现今的常规方法能取得更好的分割效果,在分割的正确率上要高于现有的常规分割方法,在处理相同数据量的点云分割时,能够达到远低于常规方法的时间消耗。通过本文提出的三维点云分割方法能够实现将大规模、复杂三维点云数据分割为较为精细、具有准确模型参数的三维几何图元,为后续实现大规模、复杂场景的精确三维构象提供有力支持。     

一种基于点云数据的建筑物平面精细分割方法

针对现有大规模点云数据平面特征分割方法中存在的错误识别、效率低、抗噪性差等问题,该文提出一种基于2D霍夫变换和八叉树的建筑物平面精细分割方法。该方法首先,对原始点云进行空间均匀降采样并向X-Y面投影,利用改进的2D霍夫变换算法提取投影后的点云线段,使用选权迭代法精确计算线段所在直线的方程及端点坐标,进一步确定立面的空间几何方程;接下来,建立原始点云数据的八叉树结构,利用端点坐标设计立方体并分割出立方体内的立面点云;最后,将立面点云从原始点云中剔除,对余下点云降采样并向X-Z面投影,重复以上过程分割水平面点云。试验验证了该文方法对建筑物面状特征分割的有效性。     

An improved segmentation approach for planar surfaces from unstructured 3D point clouds

The extraction of object features from massive unstructured point clouds with different local densities, especially in the presence of random noisy points, is not a trivial task even if that feature is a planar surface. Segmentation is the most important step in the feature extraction process. In practice, most segmentation approaches use geometrical information to segment the 3D point cloud. The features generally include the position of each point (X, Y and Z), locally estimated surface normals and residuals of best fitting surfaces; however, these features could be affected by noisy points and in consequence directly affect the segmentation results. Therefore, massive unstructured and noisy point clouds also lead to bad segmentation (over-segmentation, under-segmentation or no segmentation). While the RANSAC (random sample consensus) algorithm is effective in the presence of noise and outliers, it has two significant disadvantages, namely, its efficiency and the fact that the plane detected by RANSAC may not necessarily belong to the same object surface; that is, spurious surfaces may appear, especially in the case of parallel-gradual planar surfaces such as stairs. The innovative idea proposed in this paper is a modification for the RANSAC algorithm called Seq-NV-RANSAC. This algorithm checks the normal vector (NV) between the existing point clouds and the hypothesised RANSAC plane, which is created by three random points, under an intuitive threshold value. After extracting the first plane, this process is repeated sequentially (Seq) and automatically, until no planar surfaces can be extracted from the remaining points under the existing threshold value. This prevents the extraction of spurious surfaces, brings an improvement in quality to the computed attributes and increases the degree of automation of surface extraction. Thus the best fit is achieved for the real existing surfaces.     

Quality Assessment Of Digital Surface Models Generated From IKONOS Imagery

The growing applications of digital surface models (DSMs) for object detection, segmentation and representation of terrestrial landscapes have provided impetus for further automation of 3D spatial information extraction processes. While new technologies such as lidar are available for almost instant DSM generation, the use of stereoscopic high-resolution satellite imagery (HRSI), coupled with image matching, affords cost-effective measurement of surface topography over large coverage areas. This investigation explores the potential of IKONOS Geo stereo imagery for producing DSMs using an alternative sensor orientation model, namely bias-corrected rational polynomial coefficients (RPCs), and a hybrid image-matching algorithm. To serve both as a reference surface and a basis for comparison, a lidar DSM was employed in the Hobart testfield, a region of differing terrain types and slope. In order to take topographic variation within the modelled surface into account, the lidar strip was divided into separate sub-areas representing differing land cover types. It is shown that over topographically diverse areas, heighting accuracy to better than 3 pixels can be readily achieved. Results improve markedly in feature-rich open and relatively flat terrain, with sub-pixel accuracy being achieved at check points surveyed using the global positioning system (GPS). This assessment demonstrates that the outlook for DSM generation from HRSI is very promising.