3D Voxel Grid Research Articles

Robust and smooth Couinaud segmentation via anatomical structure-guided point-voxel network

Precise Couinaud segmentation from preoperative liver computed tomography (CT) is crucial for surgical planning and lesion examination. However, this task is challenging as it is defined based on vessel structures, and there is no intensity contrast between adjacent Couinaud segments in CT images. To solve this challenge, we design a multi-scale point-voxel fusion framework, which can more effectively model the spatial relationship of points and the semantic information of the image, producing robust and smooth Couinaud segmentations. Specifically, we first segment the liver and vessels from the CT image and generate 3D liver point clouds and voxel grids embedded with the vessel structure. Then, our method with two input-specific branches extracts complementary feature representations from points and voxels, respectively. The local attention module adaptively fuses features from the two branches at different scales to balance the contribution of different branches in learning more discriminative features. Furthermore, we propose a novel distance loss at the feature level to make the features in the segment more compact, thereby improving the certainty of segmentation between segments. Our experimental results on three public liver datasets demonstrate that our proposed method outperforms several state-of-the-art methods by large margins. Specifically, in out-of-distribution (OOD) testing of LiTS dataset, our method exceeded the voxel-based 3D UNet by approximately 20% in Dice score, and outperformed the point-based PointNet2Plus by approximately 8% in Dice score. Our code and manual annotations of the public datasets presented in this paper are available online: https://github.com/xukun-zhang/Couinaud-Segmentation.

Generalizing feature preservation in iso-surface extraction from triple dexel models

We present a method to resolve visual artifacts of a state-of-the-art iso-surface extraction algorithm by generating feature-preserving surface patches for isolated arbitrarily complex, single voxels without the need for further adaptive subdivision. In the literature, iso-surface extraction from a 3D voxel grid is limited to a single sharp feature per minimal unit, even for algorithms such as Cubical Marching Squares that produce feature-preserving surface reconstructions. In practice though, multiple sharp features can meet in a single voxel. This is reflected in the triple dexel model, which is used in simulation of CNC manufacturing processes. Our approach generalizes the use of normal information to perfectly preserve multiple sharp features for a single voxel, thus avoiding visual artifacts caused by state-of-the-art procedures.

3DWS: reliable segmentation on intelligent welding systems with 3D convolutions

AbstractAutomated industrial welding processes depend on a large number of factors interacting with high complexity resulting in some sporadic and random variability of the manufactured product that may affect its quality. It is therefore very important to have an accurate and stable quality control. In this work, a deep learning (DL) model is developed for semantic segmentation of weld seams using 3D stereo images of the seam. The objective is to correctly identify the shape and volume of the weld seam as this is the basic problem of quality control. To achieve this, a model called UNet$${\mathcal {L}}$$ L ++ has been developed, based on the UNet and UNet++ architectures, with a more complex topology and a simple encoder to achieve a good adaptation to the specific characteristics of the 3D data. The proposed model receives as input a voxelized 3D point cloud of the freshly welded part where noise is abundantly visible, and generates as output another 3D voxel grid where each voxel is semantically labeled. The experiments performed with parts built by a real weld line show a correct identification of the weld seams, obtaining values between 0.935 and 0.941 for the Dice Similarity Coefficient (DSC). As far as the authors are aware, this is the first 3D analysis proposal capable of generating shape and volume information of weld seams with almost perfect noise filtering.

Understanding structure-guided variant effect predictions using 3D convolutional neural networks.

Predicting pathogenicity of missense variants in molecular diagnostics remains a challenge despite the available wealth of data, such as evolutionary information, and the wealth of tools to integrate that data. We describe DeepRank-Mut, a configurable framework designed to extract and learn from physicochemically relevant features of amino acids surrounding missense variants in 3D space. For each variant, various atomic and residue-level features are extracted from its structural environment, including sequence conservation scores of the surrounding amino acids, and stored in multi-channel 3D voxel grids which are then used to train a 3D convolutional neural network (3D-CNN). The resultant model gives a probabilistic estimate of whether a given input variant is disease-causing or benign. We find that the performance of our 3D-CNN model, on independent test datasets, is comparable to other widely used resources which also combine sequence and structural features. Based on the 10-fold cross-validation experiments, we achieve an average accuracy of 0.77 on the independent test datasets. We discuss the contribution of the variant neighborhood in the model's predictive power, in addition to the impact of individual features on the model's performance. Two key features: evolutionary information of residues in the variant neighborhood and their solvent accessibilities were observed to influence the predictions. We also highlight how predictions are impacted by the underlying disease mechanisms of missense mutations and offer insights into understanding these to improve pathogenicity predictions. Our study presents aspects to take into consideration when adopting deep learning approaches for protein structure-guided pathogenicity predictions.

Multi-scale latent feature-aware network for logical partition based 3D voxel reconstruction

Although prior methods have achieved promising performance for recovering the 3D geometry from a single depth image, they tend to produce incomplete 3D shapes with noise. To this end, we propose Multi-Scale Latent Feature-Aware Network (MLANet) to recover the full 3D voxel grid from a single depth view of an object. MLANet logically represents a 3D voxel grid as visible voxels, occluded voxels and non-object voxels, and aims to the reconstruction of the latter two. Thus MLANet first introduces Multi-Scale Latent Feature-Aware (MLFA) based AutoEncoder (MLFA-AE) and a logical partition module to predict an occluded voxel grid (OccVoxGd) and a non-object voxel grid (NonVoxGd) from the visible voxel grid (VisVoxGd) corresponding to the input. MLANet then introduces MLFA based Generative Adversarial Network (MLFA-GAN) to refine the OccVoxGd and the NonVoxGd, and combines them with the VisVoxGd to generate a target 3D occupancy grid. MLFA shows a strong ability of learning multi-scale features of an object effectively and can be considered as a plug-and-play component to promote existing networks. The logical partition helps suppress NonVoxGd noise and improve OccVoxGd accuracy under adversarial constraints. Experimental studies on both synthetic and real-world data show that MLANet outperforms the state-of-the-art methods, and especially reconstructs unseen object categories with a higher accuracy.

GPU-accelerated Monte Carlo simulation of electron and photon interactions for radiotherapy applications

Objective. The Monte Carlo simulation software is a valuable tool in radiation therapy, in particular to achieve the needed accuracy in the dose evaluation for the treatment plans optimisation. The current challenge in this field is the time reduction to open the way to many clinical applications for which the computational time is an issue. In this manuscript we present an innovative GPU-accelerated Monte Carlo software for dose valuation in electron and photon based radiotherapy, developed as an update of the FRED (Fast paRticle thErapy Dose evaluator) software. Approach. The code transports particles through a 3D voxel grid, while scoring their energy deposition along their trajectory. The models of electromagnetic interactions in the energy region between 1 MeV—1 GeV available in literature have been implemented to efficiently run on GPUs, allowing to combine a fast tracking while keeping high accuracy in dose assessment. The FRED software has been bench-marked against state-of-art full MC (FLUKA, GEANT4) in the realm of two different radiotherapy applications: Intra-Operative Radio Therapy and Very High Electron Energy radiotherapy applications. Results. The single pencil beam dose-depth profiles in water as well as the dose map computed on non-homogeneous phantom agree with full-MCs at 2% level, observing a gain in processing time from 200 to 5000. Significance. Such performance allows for computing a plan with electron beams in few minutes with an accuracy of ∼%, demonstrating the FRED potential to be adopted for fast plan re-calculation in photon or electron radiotherapy applications.

Road edge detection based on combined deep learning and spatial statistics of LiDAR data

ABSTRACT Mobile laser scanning data can be used for effective extraction of road edge information, which is important in the domain of road maintenance and intelligent transportation. This paper proposes a road edge detection method that combines a deep learning and spatial statistics of point cloud data. Semantic segmentation using a deep neural network enables the effective extraction of point cloud fragments recognized as road. The process continues with the spatial statistical analysis of voxel features of data organized into a 3D voxel grid. Filtered voxels are clustered into spatially proximate clusters of similar shape, i.e. straight or curved edges.

AN EFFICIENT DEEP LEARNING APPROACH FOR GROUND POINT FILTERING IN AERIAL LASER SCANNING POINT CLOUDS

Abstract. Ground surface extraction is one of the classic tasks in airborne laser scanning (ALS) point cloud processing that is used for three-dimensional (3D) city modelling, infrastructure health monitoring, and disaster management. Many methods have been developed over the last three decades. Recently, Deep Learning (DL) has become the most dominant technique for 3D point cloud classification. DL methods used for classification can be categorized into end-to-end and non end-to-end approaches. One of the main challenges of using supervised DL approaches is getting a sufficient amount of training data. The main advantage of using a supervised non end-to-end approach is that it requires less training data. This paper introduces a novel local feature-based non end-to-end DL algorithm that generates a binary classifier for ground point filtering. It studies feature relevance, and investigates three models that are different combinations of features. This method is free from the limitations of point clouds’ irregular data structure and varying data density, which is the biggest challenge for using the elegant convolutional neural network. The new algorithm does not require transforming data into regular 3D voxel grids or any rasterization. The performance of the new method has been demonstrated through two ALS datasets covering urban environments. The method successfully labels ground and non-ground points in the presence of steep slopes and height discontinuity in the terrain. Experiments in this paper show that the algorithm achieves around 97% in both F1-score and model accuracy for ground point labelling.

Single Target SAR 3D Reconstruction Based on Deep Learning.

Synthetic aperture radar tomography (TomoSAR) is an important 3D mapping method. Traditional TomoSAR requires a large number of observation orbits however, it is hard to meet the requirement of massive orbits. While on the one hand, this is due to funding constraints, on the other hand, because the target scene is changing over time and each observation orbit consumes lots of time, the number of orbits can be fewer as required within a narrow time window. When the number of observation orbits is insufficient, the signal-to-noise ratio (SNR), peak-to-sidelobe ratio (PSR), and resolution of 3D reconstruction results will decline severely, which seriously limits the practical application of TomoSAR. In order to solve this problem, we propose to use a deep learning network to improve the resolution and SNR of 3D reconstruction results under the condition of very few observation orbits by learning the prior distribution of targets. We use all available orbits to reconstruct a high resolution target, while only very few (around 3) orbits to reconstruct a low resolution input. The low-res and high-res 3D voxel-grid pairs are used to train a 3D super-resolution (SR) CNN (convolutional neural network) model, just like ordinary 2D image SR tasks. Experiments on the Civilian Vehicle Radar dataset show that the proposed deep learning algorithm can effectively improve the reconstruction both in quality and in quantity. In addition, the model also shows good generalization performance for targets not shown in the training set.

Fast and Accurate 3D Object Detection for Lidar-Camera-Based Autonomous Vehicles Using One Shared Voxel-Based Backbone

Currently, many kinds of LiDAR-camera-based 3D object detectors have been developed with two heavy neural networks to extract view-specific features, while a LiDAR-camera-based 3D detector with only one neural network has not been implemented. To tackle this issue, this paper first presents an early-fusion method to exploit both LiDAR and camera data for fast 3D object detection with only one backbone, achieving a good balance between accuracy and efficiency. We propose a novel point feature fusion module to directly extract point-wise features from raw RGB images and fuse them with their corresponding point cloud with no backbone. In this paradigm, the backbone that extracts RGB image features is abandoned to reduce the large computation cost. Our method first voxelizes a point cloud into a 3D voxel grid and utilizes two strategies to reduce information loss during voxelization. The first strategy is to use a small voxel size (0.05m, 0.05m, 0.1m) in X-axis, Y-axis, and Z-axis, respectively, while the second one is to project the feature (e.g. intensity or height information) of point clouds onto RGB images. Numerous experiments evaluated on the KITTI benchmark suite show that the proposed approach outperforms state-of-the-art LiDAR-camera-based methods on the three classes in 3D performance (Easy, Moderate, Hard): cars (88.04%, 77.60%, 76.23%), pedestrians (66.65%, 60.49%, 54.51%), and cyclists (75.87%, 60.07%, 54.51%). Additionally, the proposed model runs at 17.8 frames per second (FPS), which is almost 2× faster than state-of-the-art fusion methods for LiDAR and camera.

Leveraging Image Analysis to Compute 3D Plant Phenotypes Based on Voxel-Grid Plant Reconstruction.

High throughput image-based plant phenotyping facilitates the extraction of morphological and biophysical traits of a large number of plants non-invasively in a relatively short time. It facilitates the computation of advanced phenotypes by considering the plant as a single object (holistic phenotypes) or its components, i.e., leaves and the stem (component phenotypes). The architectural complexity of plants increases over time due to variations in self-occlusions and phyllotaxy, i.e., arrangements of leaves around the stem. One of the central challenges to computing phenotypes from 2-dimensional (2D) single view images of plants, especially at the advanced vegetative stage in presence of self-occluding leaves, is that the information captured in 2D images is incomplete, and hence, the computed phenotypes are inaccurate. We introduce a novel algorithm to compute 3-dimensional (3D) plant phenotypes from multiview images using voxel-grid reconstruction of the plant (3DPhenoMV). The paper also presents a novel method to reliably detect and separate the individual leaves and the stem from the 3D voxel-grid of the plant using voxel overlapping consistency check and point cloud clustering techniques. To evaluate the performance of the proposed algorithm, we introduce the University of Nebraska-Lincoln 3D Plant Phenotyping Dataset (UNL-3DPPD). A generic taxonomy of 3D image-based plant phenotypes are also presented to promote 3D plant phenotyping research. A subset of these phenotypes are computed using computer vision algorithms with discussion of their significance in the context of plant science. The central contributions of the paper are (a) an algorithm for 3D voxel-grid reconstruction of maize plants at the advanced vegetative stages using images from multiple 2D views; (b) a generic taxonomy of 3D image-based plant phenotypes and a public benchmark dataset, i.e., UNL-3DPPD, to promote the development of 3D image-based plant phenotyping research; and (c) novel voxel overlapping consistency check and point cloud clustering techniques to detect and isolate individual leaves and stem of the maize plants to compute the component phenotypes. Detailed experimental analyses demonstrate the efficacy of the proposed method, and also show the potential of 3D phenotypes to explain the morphological characteristics of plants regulated by genetic and environmental interactions.

Image Collection Simulation Using High-Resolution Atmospheric Modeling

A new method is described for simulating the passive remote sensing image collection of ground targets that includes effects from atmospheric physics and dynamics at fine spatial and temporal scales. The innovation in this research is the process of combining a high-resolution weather model with image collection simulation to attempt to account for heterogeneous and high-resolution atmospheric effects on image products. The atmosphere was modeled on a 3D voxel grid by a Large-Eddy Simulation (LES) driven by forcing data constrained by local ground-based and air-based observations. The spatial scale of the atmospheric model (10–100 m) came closer than conventional weather forecast scales (10–100 km) to approaching the scale of typical commercial multispectral imagery (2 m). This approach was demonstrated through a ground truth experiment conducted at the Department of Energy Atmospheric Radiation Measurement Southern Great Plains site. In this experiment, calibrated targets (colored spectral tarps) were placed on the ground, and the scene was imaged with WorldView-3 multispectral imagery at a resolution enabling the tarps to be visible in at least 9–12 image pixels. The image collection was simulated with Digital Imaging and Remote Sensing Image Generation (DIRSIG) software, using the 3D atmosphere from the LES model to generate a high-resolution cloud mask. The high-resolution atmospheric model-predicted cloud coverage was usually within 23% of the measured cloud cover. The simulated image products were comparable to the WorldView-3 satellite imagery in terms of the variations of cloud distributions and spectral properties of the ground targets in clear-sky regions, suggesting the potential utility of the proposed modeling framework in improving simulation capabilities, as well as testing and improving the operation of image collection processes.

Automatic Extraction of High-Voltage Bundle Subconductors Using Airborne LiDAR Data

Overhead high-voltage conductors are the chief components of power lines and their safety has a strong influence on social and daily life. In the recent decade, the airborne laser scanning (ALS) technique has been widely used to capture the three-dimensional (3D) information of power lines and surrounding objects. Most of the existing methods focused on extraction of single conductors or extracted all conductors as one object class by applying machine learning techniques. Nevertheless, power line corridors (PLCs) are built with multi-loop, multi-phase structures (bundle conductors) and exist in intricate environments (e.g., mountains and forests), and thus raise challenges to process ALS data for extraction of individual conductors. This paper proposes an automated method to extract individual subconductors in bundles from complex structure of PLCs using a combined image- and point-based approach. First, the input point cloud data are grouped into 3D voxel grid and PL points and separated from pylon and tree points using the fact that pylons and trees are vertical objects while power lines are non-vertical objects. These pylons are further separated from trees by employing a statistical analysis technique and used to extract span points between two consecutive pylons; then, by using the distribution properties of power lines in each individual span, the bundles located at different height levels are extracted using image-based processing; finally, subconductors in each bundle are detected and extracted by introducing a window that slides over the individual bundle. The orthogonal plane transformation and recursive clustering procedures are exploited in each window position and a point-based processing is conducted iteratively for extraction of complete individual subconductors in each bundle. The feasibility and validity of the proposed method are verified on two Australian sites having bundle conductors in high-voltage transmission lines. Our experiments show that the proposed method achieves a reliable result by extracting the real structure of bundle conductors in power lines with correctness of 100% and 90% in the two test sites, respectively.

DLGAN: Depth-Preserving Latent Generative Adversarial Network for 3D Reconstruction

Although deep networks based methods outperform traditional 3D reconstruction methods which require multiocular images or class labels to recover the full 3D geometry, they may produce incomplete recovery and unfaithful reconstruction when facing occluded parts of 3D objects. To address these issues, we propose Depth-preserving Latent Generative Adversarial Network (DLGAN) which consists of 3D Encoder-Decoder based GAN (EDGAN, serving as a generator and a discriminator) and Extreme Learning Machine (ELM, serving as a classifier) for 3D reconstruction from a monocular depth image of an object. Firstly, EDGAN decodes a latent vector from the 2.5D voxel grid representation of an input image, and generates the initial 3D occupancy grid under common GAN losses, a latent vector loss and a depth loss. For the latent vector loss, we design 3D deep AutoEncoder (AE) to learn a target latent vector from ground truth 3D voxel grid and utilize the vector to penalize the latent vector encoded from the input 2.5D data. For the depth loss, we utilize the input 2.5D data to penalize the initial 3D voxel grid from 2.5D views. Afterwards, ELM transforms float values of the initial 3D voxel grid to binary values under a binary reconstruction loss. Experimental results show that DLGAN not only outperforms several state-of-the-art methods by a large margin on both a synthetic dataset and a real-world dataset, but also predicts more occluded parts of 3D objects accurately without class labels.

Permeability estimation on tomographic images using curved boundary schemes in the lattice Boltzmann method

The lattice Boltzmann method (LBM) is a widely-used numerical technique for simulation of single- and two-phase flow in geometries that are obtained using tomographic imaging of natural porous media. Due to ease of implementation and numerical robustness, a vast majority of LBM-based pore-scale simulations employ the so-called bounceback scheme to enforce no-slip velocity boundary condition. Bounceback, however, requires an implicit and tight coupling between the numerical (computational) and image (voxel) grid. This coupling results in large discretization errors, since the pore-matrix interface within the 3D image is rough. This leads to overestimation of the interfacial area, and thereby inaccurate permeability predictions. The use of the bounceback scheme also causes other numerical artifacts, such as viscosity-dependent permeability results. In order to address these deficiencies, in this work, the classical marching cubes algorithm is used to reconstruct a surface mesh from the 3D voxel grid; this mesh approximates the pore-matrix surface with higher accuracy compared to the inherent stair-stepped representation. In addition, (nominally) second-order accurate curved boundary schemes are used to enforce no-slip velocity conditions at the reconstructed pore-matrix interface.The various pre-processing steps, such as surface mesh generation and voxelization, that are necessary to use curved boundary schemes are described in detail. The proposed approach of using curved surfaces and boundary schemes is tested and validated on benchmark pore geometries, including a random packing of monodisperse spheres. We conclude that compared to current methods, curved boundary schemes provide a viable option for obtaining more accurate transport properties for Digital Rock Physics-based applications.

Reconstructing 3D Shapes from Multiple Sketches using Direct Shape Optimization.

3D shape reconstruction from multiple hand-drawn sketches is an intriguing way to 3D shape modeling. Currently, state-of-the-art methods employ neural networks to learn a mapping from multiple sketches from arbitrary view angles to a 3D voxel grid. Because of the cubic complexity of 3D voxel grids, however, neural networks are hard to train and limited to low resolution reconstructions, which leads to a lack of geometric detail and low accuracy. To resolve this issue, we propose to reconstruct 3D shapes from multiple sketches using direct shape optimization (DSO), which does not involve deep learning models for direct voxel-based 3D shape generation. Specifically, we first leverage a conditional generative adversarial network (CGAN) to translate each sketch into an attenuance image that captures the predicted geometry from a given viewpoint. Then, DSO minimizes a project-and-compare loss to reconstruct the 3D shape such that it matches the predicted attenuance images from the view angles of all input sketches. Based on this, we further propose a progressive update approach to handle inconsistencies among a few hand-drawn sketches for the same 3D shape. Our experimental results show that our method significantly outperforms the state-of-the-art methods under widely used benchmarks and produces intuitive results in an interactive application.

GENERATION OF GROUND TRUTH DATASETS FOR THE ANALYSIS OF 3D POINT CLOUDS IN URBAN SCENES ACQUIRED VIA DIFFERENT SENSORS

Abstract. In this work, we report a novel way of generating ground truth dataset for analyzing point cloud from different sensors and the validation of algorithms. Instead of directly labeling large amount of 3D points requiring time consuming manual work, a multi-resolution 3D voxel grid for the testing site is generated. Then, with the help of a set of basic labeled points from the reference dataset, we can generate a 3D labeled space of the entire testing site with different resolutions. Specifically, an octree-based voxel structure is applied to voxelize the annotated reference point cloud, by which all the points are organized by 3D grids of multi-resolutions. When automatically annotating the new testing point clouds, a voting based approach is adopted to the labeled points within multiple resolution voxels, in order to assign a semantic label to the 3D space represented by the voxel. Lastly, robust line- and plane-based fast registration methods are developed for aligning point clouds obtained via various sensors. Benefiting from the labeled 3D spatial information, we can easily create new annotated 3D point clouds of different sensors of the same scene directly by considering the corresponding labels of 3D space the points located, which would be convenient for the validation and evaluation of algorithms related to point cloud interpretation and semantic segmentation.

Fred: a GPU-accelerated fast-Monte Carlo code for rapid treatment plan recalculation in ion beam therapy

Ion beam therapy is a rapidly growing technique for tumor radiation therapy. Ions allow for a high dose deposition in the tumor region, while sparing the surrounding healthy tissue. For this reason, the highest possible accuracy in the calculation of dose and its spatial distribution is required in treatment planning. On one hand, commonly used treatment planning software solutions adopt a simplified beam–body interaction model by remapping pre-calculated dose distributions into a 3D water-equivalent representation of the patient morphology. On the other hand, Monte Carlo (MC) simulations, which explicitly take into account all the details in the interaction of particles with human tissues, are considered to be the most reliable tool to address the complexity of mixed field irradiation in a heterogeneous environment. However, full MC calculations are not routinely used in clinical practice because they typically demand substantial computational resources. Therefore MC simulations are usually only used to check treatment plans for a restricted number of difficult cases. The advent of general-purpose programming GPU cards prompted the development of trimmed-down MC-based dose engines which can significantly reduce the time needed to recalculate a treatment plan with respect to standard MC codes in CPU hardware. In this work, we report on the development of fred, a new MC simulation platform for treatment planning in ion beam therapy. The code can transport particles through a 3D voxel grid using a class II MC algorithm. Both primary and secondary particles are tracked and their energy deposition is scored along the trajectory. Effective models for particle–medium interaction have been implemented, balancing accuracy in dose deposition with computational cost. Currently, the most refined module is the transport of proton beams in water: single pencil beam dose–depth distributions obtained with fred agree with those produced by standard MC codes within 1–2% of the Bragg peak in the therapeutic energy range. A comparison with measurements taken at the CNAO treatment center shows that the lateral dose tails are reproduced within 2% in the field size factor test up to 20 cm. The tracing kernel can run on GPU hardware, achieving 10 million primary on a single card. This performance allows one to recalculate a proton treatment plan at 1% of the total particles in just a few minutes.

Detecting Tree Stems from Volumetric TLS Data in Forest Environments with Rich Understory

The present study introduces a method to identify tree stems from terrestrial laser scanning (TLS) data. We focused on forest environments of diverse and layered structure, which were technically characterized by strong occlusion effects with regards to laser scanning. The number and distribution of tree stems are important information for the management of protective forests against natural hazards, for forest inventory, and for ecological studies. Our approach builds upon a three-dimensional (3D) voxel grid transformation of the original point cloud data, followed by two major steps of processing. Firstly, a series of morphological operations removed leaves and branches and left only potential stem segments. Secondly, the stem segments of each tree were combined by a multipart workflow, which uses shape and neighborhood criteria. At the same time, erroneous fragments and noise were removed from the dataset. As a result, each object in the voxel grid was represented by a single connected component referring to one specific tree stem. Testing the method on nine spatially independent plots provided detection rates of 97% for the number and location of stems from mature trees with a diameter >= 12 cm and 84% for smaller trees with a minimum of 130 cm total tree height. In summary, we obtained a dataset covering the number and locations of the stems from both mature and understory trees, while not aiming at a precise reconstruction of the stem shape.

A new multi-scale 3D-GIS-approach for the assessment and dissemination of solar income of digital city models

In order to support an effective planning of solar panel constructions or passive energy measures, information of potential solar income has to consider multiple levels of scale and detail. It has to consider the regional and the detailed household scale. Recent developments in the field of digital city models and 3D GIS allow combining the advantages of both domains. We present a multi-scale 3D GIS approach for the assessment and dissemination of solar income of a digital city model. For the consideration of long and close range shadows, we combine polygonal geometries, 3D-voxel grids and 2.5D Digital Elevation Models of different resolutions in a multi-scale modelling scene. In order to allow a realistic assessment of the heterogeneous 3D shadowing patterns in urban environments, we present a texture mapping approach for homogeneous surface sampling and efficient data storage. The approach integrates the regional topographic context with detailed 3D assessments on a household scale. In order to show the capabilities of the presented approach, a performance test using different scene compilations was conducted. Comparing the performance of different geometrical representations of a city model tile in terms of computational effort and accuracy of the results, the relevance of high spatial resolutions and the consideration of full 3-dimensionality of the presented modelling approach is proven. The results are compatible to current 3D standards such as CityGML and COLLADA facilitating the dissemination of results to non-expert addressees.