Gunho Sohn

Using a Binary Space Partitioning Tree for Reconstructing Polyhedral Building Models from Airborne Lidar Data

During the past several years, point density covering topographic objects with airborne lidar (Light Detection And Ranging) technology has been greatly improved. This achievement provides an improved ability for reconstructing more complicated building roof structures; more specifically, those comprising various model primitives horizontally and/or vertically. However, the technology for automatically reconstructing such a complicated structure is thus far poorly understood and is currently based on employing a limited number of pre-specified building primitives. This paper addresses this limitation by introducing a new technique of modeling 3D building objects using a data-driven approach whereby densely collecting low-level modeling cues from lidar data are used in the modeling process. The core of the proposed method is to globally reconstruct geometric topology between adjacent linear features by adopting a BSP (Binary Space Partitioning) tree. The proposed algorithm consists of four steps: (a) detecting individual buildings from lidar data, (b) clustering laser points by height and planar similarity, (c) extracting rectilinear lines, and (d) planar partitioning and merging for the generation of polyhedral models. This paper demonstrates the efficacy of the algorithm for creating complex models of building rooftops in 3D space from airborne lidar data.

Point-based Classification of Power Line Corridor Scene Using Random Forests

The power line network, interconnecting power generation facilities and their end-users, is a critical infrastructure on which most of our socio-economic activities rely. As society becomes increasingly reliant on electricity, the rapid and effective monitoring of power line safety is critical. In particular, accurately knowing the current geometric and thermal status of power lines and identifying possible encroachments is the most important task in the power line risk management process. To facilitate this task, the correct identification of key objects comprising a power line corridor scene from remotely sensed data is the first important step. In recent years, airborne lidar has been successfully adopted as a cost-effective and accurate data source for mapping the power line corridors. However, in today’s practice, the classification of power line objects using lidar data still relies on labor-intensive data manipulation, and its automation is urgently required. To address this problem, this paper proposes a point-based supervised classification method, which enables the identification of five utility corridor objects (wires, pylons, vegetation, buildings, and low objects) using airborne lidar data. A total of 21 features were investigated to illustrate the horizontal and vertical properties of power line objects. A non-parametric discriminative classifier, Random Forests model, was trained with refined features to label raw laser point clouds. The proposed classifier showed 91.04 percent sample-weighted and 90.07 percent class-weighted classification accuracy, which indicates it could be highly valuable for large-scale, rapid compilations of corridor maps. A sensitivity analysis of the proposed classifier suggested that when compared, training with class-balanced samples improves classification performance over training with unbalanced samples, particularly with corridor objects such as wires and pylons.

Tree genera classification with geometric features from high-density airborne LiDAR

Categorical recognition of a trees genus is known to be valuable information for the effective management of forest inventories. In this paper, we present a method for learning a discriminative model using Random Forests to classify individual trees into three genera: pine, poplar, and maple. We believe that both internal and external geometric characteristics of the tree crown are related to tree form and therefore useful in classifying trees to the genus level. Our approach involves the extraction of both internal and external geometric features from a LiDAR point cloud as we believe that geometric features provide important information about the organization of the points inside the tree crown along with overall tree shape and form. We developed 24 geometric features and then reduced the number of features to increase efficiency. These geometric characteristics, computed for 160 sampled trees from eight field sites, were classified using Random Forests and achieved an 88.3% average accuracy rate by using 25% (40 trees) of the data for training.

Hybrid Ensemble Classification of Tree Genera Using Airborne LiDAR Data

Abstract: This paper presents a hybrid ensemble method that is comprised of a sequential and a parallel architecture for the classification of tree genus using LiDAR (Light Detection and Ranging) data. The two classifiers use different sets of features: (1) features derived from geometric information, and (2) features derived from vertical profiles using Random Forests as the base classifier. This classification result is also compared with that obtained by replacing the base classifier by LDA (Linear Discriminant Analysis), kNN (k Nearest Neighbor) and SVM (Support Vector Machine). The uniqueness of this research is in the development, implementation and application of three main ideas: (1) the hybrid ensemble method, which aims to improve classification accuracy, (2) a pseudo-margin criterion for assessing the quality of predictions and (3) an automatic feature reduction method using results drawn from Random Forests. An additional point-density analysis is performed to study the influence of decreased point density on classification accuracy results. By using Random Forests as the base classifier, the average classification accuracies for the geometric classifier and vertical profile classifier are 88.0% and 88.8%, respectively,

Mapping tree genera using discrete LiDAR and geometric tree metrics

Los mapas de generos de arboles son utiles para el inventario forestal, planificacion urbana y el mantenimiento de la infraestructura de lineas de transmision. Se presenta un estudio de caso de uso de datos LiDAR de alta densidad para el mapeo de generos de arboles a lo largo del derecho de paso (ROW) de un corredor de linea de transmision. El objetivo de la investigacion fue identificar arboles individuales que mostraban o poseian una amenaza potencial a la infraestructura de la linea de transmision. Mediante el uso de mapas tridimensionales de LiDAR se derivaron metricas de arboles que estan relacionadas con la geometria de estos (formas del arbol). Por ejemplo, la direccion del crecimiento dominante de los arboles es util para identificar arboles que crecen inclinados hacia las lineas de transmision. Tambien se derivaron otras metricas geometricas que son utiles para determinar los generos de los arboles, tales como altura, forma de la copa, tamano y estructura de ramas. El area de estudio se ubico al norte de Thessalon, Ontario, Canada, a lo largo de los principales corredores de ROW y en los bosques aledanos. Los atributos geometricos usados para la clasificacion de los generos fueron categorizados en cinco amplias clases: 1) lineas, 2) agrupamiento, 3) volumenes, 4) amortiguamiento en 3D de puntos, y 5) forma general del arbol que provee parametros como una entrada para el clasificador forestal aleatorio.

Maximizing the Diversity of Ensemble Random Forests for Tree Genera Classification Using High Density LiDAR Data

Recent research into improving the effectiveness of forest inventory management using airborne LiDAR data has focused on developing advanced theories in data analytics. Furthermore, supervised learning as a predictive model for classifying tree genera (and species, where possible) has been gaining popularity in order to minimize this labor-intensive task. However, bottlenecks remain that hinder the immediate adoption of supervised learning methods. With supervised classification, training samples are required for learning the parameters that govern the performance of a classifier, yet the selection of training data is often subjective and the quality of such samples is critically important. For LiDAR scanning in forest environments, the quantification of data quality is somewhat abstract, normally referring to some metric related to the completeness of individual tree crowns; however, this is not an issue that has received much attention in the literature. Intuitively the choice of training samples having varying quality will affect classification accuracy. In this paper a Diversity Index (DI) is proposed that characterizes the diversity of data quality (Qi) among selected training samples required for constructing a classification model of tree genera. The training sample is diversified in terms of data quality as opposed to the number of samples per class. The diversified training sample allows the classifier to better learn the positive and negative instances and; therefore; has a higher classification accuracy in discriminating the “unknown” class samples from the “known” samples. Our algorithm is implemented within the Random Forests base classifiers with six derived geometric features from LiDAR data. The training sample contains three tree genera (pine; poplar; and maple) and the validation samples contains four labels (pine; poplar; maple; and “unknown”). Classification accuracy improved from 72.8%; when training samples were selected randomly (with stratified sample size); to 93.8%; when samples were selected with additional criteria; and from 88.4% to 93.8% when an ensemble method was used.

Matching Aerial Images to 3D Building Models Using Context-Based Geometric Hashing

A city is a dynamic entity, which environment is continuously changing over time. Accordingly, its virtual city models also need to be regularly updated to support accurate model-based decisions for various applications, including urban planning, emergency response and autonomous navigation. A concept of continuous city modeling is to progressively reconstruct city models by accommodating their changes recognized in spatio-temporal domain, while preserving unchanged structures. A first critical step for continuous city modeling is to coherently register remotely sensed data taken at different epochs with existing building models. This paper presents a new model-to-image registration method using a context-based geometric hashing (CGH) method to align a single image with existing 3D building models. This model-to-image registration process consists of three steps: (1) feature extraction; (2) similarity measure; and matching, and (3) estimating exterior orientation parameters (EOPs) of a single image. For feature extraction, we propose two types of matching cues: edged corner features representing the saliency of building corner points with associated edges, and contextual relations among the edged corner features within an individual roof. A set of matched corners are found with given proximity measure through geometric hashing, and optimal matches are then finally determined by maximizing the matching cost encoding contextual similarity between matching candidates. Final matched corners are used for adjusting EOPs of the single airborne image by the least square method based on collinearity equations. The result shows that acceptable accuracy of EOPs of a single image can be achievable using the proposed registration approach as an alternative to a labor-intensive manual registration process.

Implicit Regularization for Reconstructing 3D Building Rooftop Models Using Airborne LiDAR Data

With rapid urbanization, highly accurate and semantically rich virtualization of building assets in 3D become more critical for supporting various applications, including urban planning, emergency response and location-based services. Many research efforts have been conducted to automatically reconstruct building models at city-scale from remotely sensed data. However, developing a fully-automated photogrammetric computer vision system enabling the massive generation of highly accurate building models still remains a challenging task. One the most challenging task for 3D building model reconstruction is to regularize the noises introduced in the boundary of building object retrieved from a raw data with lack of knowledge on its true shape. This paper proposes a data-driven modeling approach to reconstruct 3D rooftop models at city-scale from airborne laser scanning (ALS) data. The focus of the proposed method is to implicitly derive the shape regularity of 3D building rooftops from given noisy information of building boundary in a progressive manner. This study covers a full chain of 3D building modeling from low level processing to realistic 3D building rooftop modeling. In the element clustering step, building-labeled point clouds are clustered into homogeneous groups by applying height similarity and plane similarity. Based on segmented clusters, linear modeling cues including outer boundaries, intersection lines, and step lines are extracted. Topology elements among the modeling cues are recovered by the Binary Space Partitioning (BSP) technique. The regularity of the building rooftop model is achieved by an implicit regularization process in the framework of Minimum Description Length (MDL) combined with Hypothesize and Test (HAT). The parameters governing the MDL optimization are automatically estimated based on Min-Max optimization and Entropy-based weighting method. The performance of the proposed method is tested over the International Society for Photogrammetry and Remote Sensing (ISPRS) benchmark datasets. The results show that the proposed method can robustly produce accurate regularized 3D building rooftop models.

Soft classification of mixed seabed objects based on fuzzy clustering analysis using airborne LIDAR bathymetry data

Coastal seabed mapping is essential for a variety of nearshore management related activities including sustainable resource management, ecological protection, and environmental change detection in coastal sites. Recently introduced airborne LIDAR bathymetry (ALB) sensors allow, under favorable environmental conditions and mapping requirements, time and cost efficient collection of shallow coastal seabed data in comparison to acoustic techniques. One important application of these sensors, given ALB seabed footprint size on the order to several meters in diameter for shallow waters, is the development of seabed classification maps and techniques to classify both benthic species and seabed sediment. The coastal seabed is a complex environment consisting of diverse habitats and, thus, necessitates classification methods which readily account for seabed class heterogeneity. Recent ALB classification studies have relied on classification techniques that assign each ALB shot to a single seabed class (i.e., hard classification) instead of allowing for assignment to multiple seabed classes which may be present in an illuminated ALB footprint (i.e., soft classification). In this study, a soft seabed classification (SSC) algorithm is developed using unsupervised classification with fuzzy clustering to produce classification products accounting for a sub-footprint habitat mixture. With this approach, each shot is assigned to multiple seabed classes with a percentage cover measuring the extent to which each seabed class is present in the ALB footprint. This has the added benefit of generating smooth spatial ecological transitions of the seabed instead of sharp boundaries between classes or clusters. Furthermore, due to the multivariate nature of the SSC output (i.e., percentage cover for each seabed class for a given shot), a recently developed self-organizing map neural network-based approach to geo-visualization of seabed classification results was used to visualize seabed habitat diversity. An ALB dataset of an area approximately 20000 m2 collected from Quebec, Canada was used. Cross-validation of the SSC approach yields percentage cover accuracy of approximately 71.7% with 16 seabed classes for a real ALB dataset, while dominant seabed class prediction based on hardening of percentage cover predictions yielded 66% accuracy for 4 seabed classes.

Individual tree species classification using structure features from high density airborne lidar data

The paper investigated the advantage of high density airborne LiDAR data for improving species classification of individual tree. The investigation is comprised of two stages, feature extraction and classification. Several feature metrics were derived from LiDAR data, most of which were to characterize the vertical structural properties of difference species. Some other metrics were calculated statistically from intensity and return number information. A supervised decision tree algorithm was applied on the extracted features to perform both feature selection and classification. Two classification themes were carried out: classification of coniferous and deciduous trees, and classification of five species. Experiment was conducted in Canadian boreal forests dominated by mature trees. The results demonstrated LiDAR derived vertical profile metrics are capable for species classification either to separate coniferous and deciduous or to separate multiple species. The best overall classification accuracy is 81.7% validated by using the test data from the same ecosystem as the training data.