The development of tools for the generation of 3D city models started almost two decades ago. From the beginning, fully automatic reconstruction systems were envisioned to fulfil the need for efficient data collection. However, research on automatic city modelling is still a very active area. The paper will review a number of current approaches in order to comprehensively elaborate the state of the art of reconstruction methods and their respective principles. Originally, automatic city modelling only aimed at polyhedral building objects, which mainly reflects the respective roof shapes and building footprints. For this purpose, airborne images or laser scans are used. In addition to these developments, the paper will also review current approaches for the generation of more detailed facade geometries from terrestrial data collection.

An update on automatic 3D building reconstruction

Airborne lidar systems have become a source for the acquisition of elevation data. They provide georeferenced, irregularly distributed 3D point clouds of high altimetric accuracy. Moreover, these systems can provide for a single laser pulse, multiple returns or echoes, which correspond to different illuminated objects. In addition to multi-echo laser scanners, full-waveform systems are able to record 1D signals representing a train of echoes caused by reflections at different targets. These systems provide more information about the structure and the physical characteristics of the targets. Many approaches have been developed, for urban mapping, based on aerial lidar solely or combined with multispectral image data. However, they have not assessed the importance of input features. In this paper, we focus on a multi-source framework using aerial lidar (multi-echo and full waveform) and aerial multispectral image data. We aim to study the feature relevance for dense urban scenes. The Random Forests algorithm is chosen as a classifier: it runs efficiently on large datasets, and provides measures of feature importance for each class. The margin theory is used as a confidence measure of the classifier, and to confirm the relevance of input features for urban classification. The quantitative results confirm the importance of the joint use of optical multispectral and lidar data. Moreover, the relevance of full-waveform lidar features is demonstrated for building and vegetation area discrimination.

http://recherche.ign.fr/labos/matis/pdf/articles_revues/chehata_etal_IJPRS10.pdf

Relevance of airborne lidar and multispectral image data for urban scene classification using Random Forests

Small-footprint (0.2–2 m) airborne laser scanners are lidar instruments originally developed for topographic mapping. While the first airborne laser scanners only allowed determining the range from the sensor to the target, the latest sensor generation records the complete echo waveform. The waveform provides important information about the backscattering properties of the observed targets and may be useful for geophysical parameter retrieval and advanced geometric modelling. However, to fully utilise the potential of the waveform measurements in applications, it is necessary to perform a radiometric calibration. As there are not yet calibration standards, this paper reviews some basic physical concepts commonly used by the remote sensing community for modelling scattering and reflection processes. Based purely on theoretical arguments it is recommended to use the backscattering coefficient γ , which is the backscatter cross-section normalised relative to the laser footprint area, for the radiometric calibration of small-footprint full-waveform airborne laser scanners. The presented concepts are, with some limitations, also applicable to conventional airborne laser scanners that measure the range and intensity of multiple echoes.

Radiometric calibration of small-footprint full-waveform airborne laser scanner measurements: Basic physical concepts

Object-based point cloud analysis (OBPA) is useful for information extraction from airborne LiDAR point clouds. An object-based classification method is proposed for classifying the airborne LiDAR point clouds in urban areas herein. In the process of classification, the surface growing algorithm is employed to make clustering of the point clouds without outliers, thirteen features of the geometry, radiometry, topology and echo characteristics are calculated, a support vector machine (SVM) is utilized to classify the segments, and connected component analysis for 3D point clouds is proposed to optimize the original classification results. Three datasets with different point densities and complexities are employed to test our method. Experiments suggest that the proposed method is capable of making a classification of the urban point clouds with the overall classification accuracy larger than 92.34% and the Kappa coefficient larger than 0.8638, and the classification accuracy is promoted with the increasing of the point density, which is meaningful for various types of applications.

SVM-Based Classification of Segmented Airborne LiDAR Point Clouds in Urban Areas

Ripe with possibilities offered by deep-learning techniques and useful in applications related to remote sensing, computer vision, and robotics, 3D point cloud semantic segmentation (PCSS) and point cloud segmentation (PCS) are attracting increasing interest. This article summarizes available data sets and relevant studies on recent developments in PCSS and PCS.

/pdf/linking-points-with-labels-in-3d-a-review-of-point-cloud-1venutor5l.pdf

Linking Points With Labels in 3D: A Review of Point Cloud Semantic Segmentation

Abstract Segmentation is a key task in the processing of 3D point clouds as obtained from airborne laser scanners (ALS). However, most of the segmentation techniques currently employed require prior gridding of the data and thus do not respect the inherently three-dimensional geometry of more intricate structures such as power lines. By contrast, the mean shift procedure, a filtering and clustering approach which has recently found much interest in the image processing community, works directly on the original 3D point cloud; also, mean shift is a non-parametric technique (i.e., it does not depend on any geometric model assumptions) and can thus also be applied to vegetation structures. In this paper, we will give a self-contained derivation of the mean shift procedure, and discuss how it can be used to obtain a classification or segmentation of an unstructured 3D point cloud. Two application examples shall further illustrate its usefulness to ALS data processing.

Thomas Melzer

Papers

Non-parametric segmentation of ALS point clouds using mean shift