When users zoom in or out on a digital map, the map should change correspondingly to present geographical information at proper levels. A way to help map users better keep track of their interested objects is to change the map smoothly instead of discretely switching between several levels of detail. This paper focuses on the problem of providing smooth merging of area objects. We propose to merge multiple areas simultaneously to share their animation durations. In this way, each merging operation can be prolonged, and it is visually smoother. We present a greedy algorithm to decide which areas should be merged at each step. The merging process is pre-computed and is recorded into a space-scale cube (SSC). When a user accesses our web map, the SSC is sent to the client side so that the map can be generated by slicing the SSC in the graphics processing unit (GPU). We also explain how to snap the zooming to valid states so that the zooming will not stop halfway of the merging operations. Our case study shows that it is visually smoother to merge simultaneously than to sequentially merge each pair of areas.

We investigate how PCServe, a web service for disseminating massive point clouds, performs for read-only access (i.e. a visualization application). PCServe is backed by a database model based on Space Filling Curves. By adding a virtual hierarchy of blocks to the database, we can support different visualization applications for retrieval of point cloud data over the web without having to store the data multiple times. This makes expressive access to point clouds over the web possible. We investigate the amount of processing that is needed to create the database model and how well PCServe handles requests from the visualization application. Some suggestions are provided how the current approach can be improved.

Point clouds contain high detail and high accuracy geometry representation of the scanned Earth surface parts. To manage the huge amount of data, the point clouds are traditionally organized on location and map-scale; e.g. in an octree structure, where top-levels of the tree contain few points suitable for small scale overviews and lower levels of the tree contain more points suitable for large scale detailed views. The drawback of this solution is that it is based on discrete levels, causing visual artifacts in the form of data density shocks when creating the commonly used perspective views. This paper presents a method based on an optimized distribution of points over continuous levels, avoiding the visualization shocks. The traditional distribution ratio's of data amounts over discrete levels of raster or vector data is considered the reference. How to convert this to point clouds with continuous levels (still benefiting from the proven advantages of the data distribution in discrete levels for efficient access at a wide range of scales)? In our solution, for each point a cLoD (continuous Level of Detail) value is computed and added as dimension to the point. A SFC (Space Filling Curve)-based nD data clustering technique can be used to organize the points, so that they can be efficiently queried. It should be noted that also other multi-dimensional indexing and clustering techniques could be applied to realize continuous levels based on the cLoD value. Besides the mathematical foundation of the approach also several implementations are described, varying from a 3D web-browser based solution to an augmented reality point cloud app in a mobile phone. The cLoD enables interactive real-time visualization using perspective views without data density shocks, while supporting continuous zoom-in/out and progressive data streaming between server and client. The described cLoD based approach is generic and supports different types of point clouds: from airborne, terrestrial, mobile and indoor laser scanning, but also from dense matching optical imagery or multi-beam echo soundings.

Efficient spatial queries are frequently needed to extract useful information from massive nD point clouds. Most previous studies focus on developing solutions for orthogonal window queries, while rarely considering the polytope query. The latter query, which includes the widely adopted polygonal query in 2D, also plays a critical role in many nD spatial applications such as the perspective view selection. Aiming for an nD solution, this paper first formulates a convex nD-polytope for querying. Then, the paper integrates three approximate geometric algorithms – SWEEP, SPHERE, VERTEX, and a linear programming method CPLEX, developing a solution based on an Index-Organized Table (IOT) approach. IOT is applied with space filling curve based clustering and advanced querying mechanism which recursively refines hypercubic nD spaces to approach the query geometry for primary filtering. Results from experiments based on both synthetic and real data have confirmed the superior performance of SWEEP. However, the algorithm may lag behind CPLEX due to pessimistic intersection computation in high dimensional spaces. In a real application, by properly transforming a perspective view selection into a polytope query, the solution achieves a sub-second querying performance using SWEEP. In another flood risk query, SWEEP also leads the others. In general, the robust and efficient solution can be immediately used to address different polytope queries, including those abstract ones whose constraints on combinations of different dimensions are formed into a polytope model. Besides, the knowledge of high-dimensional computations acquired also provides significant guidance for handling more nD GIS issues.

Point cloud data have rich semantic representations and can benefit various applications towards a digital twin. However, they are unordered and anisotropically distributed, thus being unsuitable for a typical Convolutional Neural Networks (CNN) to handle. With the advance of deep learning, several neural networks claim to have solved the point cloud semantic segmentation problem. This paper evaluates three different neural networks for semantic segmentation of point clouds, namely PointNet++, PointCNN and DGCNN. A public indoor scene of the Amersfoort railway station is used as the study area. Unlike the typical indoor scenes and even more from the ubiquitous outdoor ones in currently available datasets, the station consists of objects such as the entrance gates, ticket machines, couches, and garbage cans. For the experiment, we use subsets from the data, remove the noise, evaluate the performance of the selected neural networks. The results indicate an overall accuracy of more than 90% for all the networks but vary in terms of mean class accuracy and mean Intersection over Union (IoU). The misclassification mainly occurs in the classes of couch and garbage can. Several factors that may contribute to the errors are analyzed, such as the quality of the data and the proportion of the number of points per class. The adaptability of the networks is also heavily dependent on the training location: the overall characteristics of the train station make a trained network for one location less suitable for another.

Governments use flood maps for city planning and disaster management to protect people and assets. Flood risk mapping projects carried out for these purposes generate a huge amount of modelling results. Previously, data submitted are highly condensed products such as typical flood inundation maps and tables for loss analysis. Original modelling results recording critical flood evolution processes are overlooked due to cumbersome management and analysis. This certainly has drawbacks: the ĝ€ static' maps impart few details about the flood; also, the data fails to address new requirements. This significantly confines the use of flood maps. Recent development of point cloud databases provides an opportunity to manage the whole set of modelling results. The databases can efficiently support all kinds of flood risk queries at finer scales. Using a case study from China, this paper demonstrates how a novel nD-PointCloud structure, HistSFC, improves flood risk querying. The result indicates that compared with conventional database solutions, HistSFC holds superior performance and better scalability. Besides, the specific optimizations made on HistSFC can facilitate the process further. All these indicate a promising solution for the next generation of flood maps.