GSGeorge Sithole

George Sithole

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


George Sithole has a total of 22 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Spaces in Spatial Science and Urban Applications—State of the Art Review

    by Sisi Zlatanova, Jinjin Yan, Yijing Wang, Abdoulaye Diakité, Umit Isikdag, George Sithole, Jack Barton

    In spatial science and urban applications, “space" is presented by multiple disciplines as a notion referencing our living environment. Space is used as a general term to help understand particular characteristics of the environment. However, the definition and perception of space varies and these variations have to be harmonised. For example, space may have diverse definitions and classification, the same environment may be abstracted/modelled by contradicting notions of space, which can lead to inconsistencies and misunderstandings. In this paper, we seek to investigate and document the state-of-the-art in the research of “space” regarding its definition, classification, modelling and utilization (2D/3D) in spatial sciences and urban applications. We focus on positioning, navigation, building micro-climate and thermal comfort, landscape, urban planning and design, urban heat island, interior design and planning, transportation and intelligent space. We review 147 research papers, technical reports and on-line resources. We compare the presented space concepts with respect to five criteria—classification, boundary, modelling components, use of standards and granularity. The review inventory is intended for both scientists and professionals in the spatial industry, such as companies, national mapping agencies and governments, and aim to provide a reference to better understand and employ the “space” while working across disciplines.


  • Open Access

    AUTOMATICALLY DETERMINING SCALE WITHIN UNSTRUCTURED POINT CLOUDS

    by Jayren Kadamen, George Sithole

    Three dimensional models obtained from imagery have an arbitrary scale and therefore have to be scaled. Automatically scaling these models requires the detection of objects in these models which can be computationally intensive. Real-time object detection may pose problems for applications such as indoor navigation. This investigation poses the idea that relational cues, specifically height ratios, within indoor environments may offer an easier means to obtain scales for models created using imagery. The investigation aimed to show two things, (a) that the size of objects, especially the height off ground is consistent within an environment, and (b) that based on this consistency, objects can be identified and their general size used to scale a model. To test the idea a hypothesis is first tested on a terrestrial lidar scan of an indoor environment. Later as a proof of concept the same test is applied to a model created using imagery. The most notable finding was that the detection of objects can be more readily done by studying the ratio between the dimensions of objects that have their dimensions defined by human physiology. For example the dimensions of desks and chairs are related to the height of an average person. In the test, the difference between generalised and actual dimensions of objects were assessed. A maximum difference of 3.96% (2.93cm) was observed from automated scaling. By analysing the ratio between the heights (distance from the floor) of the tops of objects in a room, identification was also achieved.

  • Open Access

    POSITION, LOCATION, PLACE AND AREA: AN INDOOR PERSPECTIVE

    by George Sithole, Sisi Zlatanova

    Over the last decade, harnessing the commercial potential of smart mobile devices in indoor environments has spurred interest in indoor mapping and navigation. Users experience indoor environments differently. For this reason navigational models have to be designed to adapt to a user’s personality, and to reflect as many cognitive maps as possible. This paper presents an extension of a previously proposed framework. In this extension the notion of placement is accounted for, thereby enabling one aspect of the ‘personalised indoor experience’. In the paper, firstly referential expressions are used as a tool to discuss the different ways of thinking of placement within indoor spaces. Next, placement is expressed in terms of the concept of Position, Location, Place and Area. Finally, the previously proposed framework is extended to include these concepts of placement. An example is provided of the use of the extended framework. Notable characteristics of the framework are: (1) Sub-spaces, resources and agents can simultaneously possess different types of placement, e.g., a person in a room can have an xyz position and a location defined by the room number. While these entities can simultaneously have different forms of placement, only one is dominant. (2) Sub-spaces, resources and agents are capable of possessing modifiers that alter their access and usage. (3) Sub-spaces inherit the modifiers of the resources or agents contained in them. (4) Unlike conventional navigational models which treat resources and obstacles as different types of entities, in the proposed framework there are only resources and whether a resource is an obstacle is determined by a modifier that determines whether a user can access the resource. The power of the framework is that it blends the geometry and topology of space, the influence of human activity within sub-spaces together with the different notions of placement in a way that is simple and yet very flexible.


  • Open Access

    FRAMEWORK FOR COMPARING SEGMENTATION ALGORITHMS

    by G. Sithole, L. Majola

    The notion of a ‘Best’ segmentation does not exist. A segmentation algorithm is chosen based on the features it yields, the properties of the segments (point sets) it generates, and the complexity of its algorithm. The segmentation is then assessed based on a variety of metrics such as homogeneity, heterogeneity, fragmentation, etc. Even after an algorithm is chosen its performance is still uncertain because the landscape/scenarios represented in a point cloud have a strong influence on the eventual segmentation. Thus selecting an appropriate segmentation algorithm is a process of trial and error. Automating the selection of segmentation algorithms and their parameters first requires methods to evaluate segmentations. Three common approaches for evaluating segmentation algorithms are ‘goodness methods’, ‘discrepancy methods’ and ‘benchmarks’. Benchmarks are considered the most comprehensive method of evaluation. This paper shortcomings in current benchmark methods are identified and a framework is proposed that permits both a visual and numerical evaluation of segmentations for different algorithms, algorithm parameters and evaluation metrics. The concept of the framework is demonstrated on a real point cloud. Current results are promising and suggest that it can be used to predict the performance of segmentation algorithms.

  • Open Access

    Enhancing Manual Scan Registration Using Audio Cues

    by T. Ntsoko, G. Sithole

    Indoor mapping and modelling requires that acquired data be processed by editing, fusing, formatting the data, amongst other operations. Currently the manual interaction the user has with the point cloud (data) while processing it is visual. Visual interaction does have limitations, however. One way of dealing with these limitations is to augment audio in point cloud processing. Audio augmentation entails associating points of interest in the point cloud with audio objects. In coarse scan registration, reverberation, intensity and frequency audio cues were exploited to help the user estimate depth and occupancy of space of points of interest. Depth estimations were made reliably well when intensity and frequency were both used as depth cues. Coarse changes of depth could be estimated in this manner. The depth between surfaces can therefore be estimated with the aid of the audio objects. Sound reflections of an audio object provided reliable information of the object surroundings in some instances. For a point/area of interest in the point cloud, these reflections can be used to determine the unseen events around that point/area of interest. Other processing techniques could benefit from this while other information is estimated using other audio cues like binaural cues and Head Related Transfer Functions. These other cues could be used in position estimations of audio objects to aid in problems such as indoor navigation problems.

  • Open Access

    Problems In Indoor Mapping and Modelling

    by S. Zlatanova, G. Sithole, M. Nakagawa, Q. Zhu

    Research in support of indoor mapping and modelling (IMM) has been active for over thirty years. This research has come in the form of As-Built surveys, Data structuring, Visualisation techniques, Navigation models and so forth. Much of this research is founded on advancements in photogrammetry, computer vision and image analysis, computer graphics, robotics, laser scanning and many others. While IMM used to be the privy of engineers, planners, consultants, contractors, and designers, this is no longer the case as commercial enterprises and individuals are also beginning to apply indoor models in their business process and applications. There are three main reasons for this. Firstly, the last two decades have seen greater use of spatial information by enterprises and the public. Secondly, IMM has been complimented by advancements in mobile computing and internet communications, making it easier than ever to access and interact with spatial information. Thirdly, indoor modelling has been advanced geometrically and semantically, opening doors for developing user-oriented, context-aware applications. This reshaping of the public’s attitude and expectations with regards to spatial information has realised new applications and spurred demand for indoor models and the tools to use them. This paper examines the present state of IMM and considers the research areas that deserve attention in the future. In particular the paper considers problems in IMM that are relevant to commercial enterprises and the general public, groups this paper expects will emerge as the greatest users IMM. The subject of indoor modelling and mapping is discussed here in terms of Acquisitions and Sensors, Data Structures and Modelling, Visualisation, Applications, Legal Issues and Standards. Problems are discussed in terms of those that exist and those that are emerging. Existing problems are those that are currently being researched. Emerging problems are those problems or demands that are expected to arise because of social changes, technological advancements, or commercial interests. The motivation of this work is to define a set of research problems that are either being investigated or should be investigated. These will hopefully provide a framework for assessing progress and advances in indoor modelling. The framework will be developed in the form of a problem matrix, detailing existing and emerging problems, their solutions and present best practices. Once the framework is complete it will be published online so that the IMM community can discuss and modify it as necessary. When the framework has reached a steady state an empirical benchmark will be provided to test solutions to posed problems. A yearly evaluation of the problem matrix will follow, the results of which will be published.


  • Open Access

    ASSESSMENT OF THE HOMOGENEITY OF VOLUNTEERED GEOGRAPHIC INFORMATION IN SOUTH AFRICA

    by L. Siebritz, G. Sithole, S. Zlatanova

    The potential for volunteer groups to contribute geographic data to National Mapping Agencies has been widely recognised. Several investigations have been done to determine the geometric accuracy of this data for the purposes of national mapping. Beyond accuracy, from a production perspective National Mapping Agencies will also be interested in the sufficiency and uniformity of the data. This paper presents an investigation of whether presently geographic data generated by volunteers is uniform across a country and whether the rate of production of data is consistent. For the purpose of the test, changes in data of South Africa from OpenStreetMap are analysed for the period 2006 to 2011. Here only point and line data are considered. The results generally show that the rate at which data is generated varies in space and time. The results also confirm that volunteers emphasise on the capture of certain information and that the capture doesn’t average out as might be expected. The results also showed that social events, such as a World Cup, also have the effect of spurring the generation of volunteer geographic data. The implication of these results for National Mapping Agencies is that they cannot treat volunteer geographic information as being of a uniform standard. How National Mapping Agencies respond to this will have to be the subject of other investigations.


  • Open Access

    MOVING WINDOW SEGMENTATION FRAMEWORK FOR POINT CLOUDS

    by G. Sithole, B. Gorte

    As lidar point clouds become larger streamed processing becomes more attractive. This paper presents a framework for the streamed segmentation of point clouds with the intention of segmenting unstructured point clouds in real-time. The framework is composed of two main components. The first component segments points within a window shifting over the point cloud. The second component stitches the segments within the windows together. In this fashion a point cloud can be streamed through these two components in sequence, thus producing a segmentation. The algorithm has been tested on airborne lidar point cloud and some results of the performance of the framework are presented.

  • Open Access

    DEFORMATION DETECTION IN PIPING INSTALLATIONS USING PROFILING TECHNIQUES

    by W. T. Mapurisa, G. Sithole

    As-Built surveys of buildings and installations have received greater attention in recent years. An example is the 3D digital reconstruction of existing piping installations at chemical plants for plant maintenance, upgrades and safety standard concerns. This paper is directed at the reconstruction of piping installations with particular emphasis on the detection of deformities in pipes. Reconstruction begins with the automatic detection of individual piping elements which requires a prior segmentation. For segmentation, the profile intersection technique is used. Surfaces are considered as a network of intersecting curves as opposed to surface patches. Recreating such curves on a point set, and intersecting them, segments are identified. The entire scan is partitioned into a series of scan planes referred to as profiles. Points are then connected in each profile based on the surface they represent forming line segments. The line segments, which represent curves, are then intersected to identify segments. For pipes, line segments are elliptical. The centre of an ellipse lies on the pipes’ axis and the semi minor axis is equivalent to the radius of the pipe. Therefore together the centres and semi minor axis are used to describe the position, orientation, size and radius of a pipe. For deformed pipes, the line segments deviate from the elliptical shape. By identifying deviations of the line segments from the elliptical shape deformations are identified. The algorithm allows for cylinders, spheres, cones and tori to be detected including deformities in their shape. Experimental results show the effectiveness of the algorithm.