Skip to content

HTTPS clone URL

Subversion checkout URL

You can clone with
or
.
Download ZIP
Fetching contributors…

Cannot retrieve contributors at this time

237 lines (169 sloc) 7.362 kB

Display and Management of Geomatics Research Data

.fx: title

Display and Management of Geomatics Research Data

Timothy Daniel Trewartha and Michiel Johan Baird

oct.png

Zamani Project

  • Started by the UCT Department of Geomatics in 2004
  • Aims to preserve cultural heritage by documenting sites and producing laser scanned models
  • Over 100 models of sites in various African countries including South Africa, Zimbabwe, Kenya and Tanzania
  • Some of the models are very large and contain billions of points
zamani.jpg

Problems Faced

  • Fast-growing volume of data
  • Difficulty in storing the data
  • Difficulty with viewing the large models in real-time
  • Data locality issues
  • Large number of users interact with the data

Solution Approach

  • To enable viewing of the large 3D models at interactive frame rates we will implement a hierachical multi-resolution data structure
  • A server with a large amount of storage will be procured; this will be used to store the models and GIS data, and support the core functionalities of the solution.

Dynamic Viewing of Large 3D Models - Tim

  • The Department of Geomatics has indicated that they have difficulties handling the sizes of some of their models
  • Some of the models they are dealing with contain over 8 billion points
  • At this point, traditional viewing methods cannot cope; the resolution of the original model must be decreased manually beforehand
  • This compromise is often unacceptable
greatzimbabwe.jpg

Research Question

  • Is it feasible to support real time viewing of models containing billions of points?
  • Answering this question in the affirmative would enable exploration of the Zamani models in their full detail
  • It would have a significant impact in the Geomatics department

Proposed Solution

  • Implement a multi-resolution data structure to divide our model into manageable chunks
  • Initially only a subset of the points is available
  • As one zooms into the model and greater detail is required, we dynamically fetch additional points from our data structure until the full original detail is available

Relevant Literature

  • Common approaches to structuring large 3D models include octrees, R-trees, bounding sphere hierachies, and Hilbert Space Filling Curves
Data Structure Largest model rendered Reference
Bounding Sphere Hierachy 8 million points (Rusinkiewicz and Levoy, 2000)
Hierarchy of Tetrahedra 300 million points (Cigoni et al., 2008)
Octree 2.2 billion points (Wand et al., 2007)

Proposed Datastructure

  • From researching the literature it seems that octrees have the best performance
  • All data is stored in the leaf nodes
  • Inner nodes provide simplified multi-resolution representations
  • No leaf node should contain more than a specified number of points
  • Empirically, it seems that a value of around 30,000 gives good performance
oct.png

Evaluation Criteria

  • Can the system render the largest of the Zamani models at interactive frame rates?
  • If this goal is achieved the system will be a success
  • Varying degrees of success can also be determined by testing smaller models of varying sizes

Workbench - Michiel

How effective is an automated workflow solution in GIS context?

workflow.png

Previous work

  • Various fields of science has adopted and implemented workflow systems
  • These systems have increased efficiency and research output
  • GIS research has been shown to be applicable to an automated workflow system; this has not however been implemented

Proposed Solution

  • Use an existing workflow system as various platforms already exist
  • Design a workflow that is applicable to GIS
  • Write middleware to integrate with existing GIS tools
  • Software that automatically transfers data as it is needed down the pipeline

Testing Criteria

  • How much does the content delivery system decrease waiting time?
  • How effective is the workflow system based on the analytics that will be generated be the system.

Division of Work

  • Tim will be implementing the hierachical datastructure to enable real-time interaction with the large 3D models.
  • Michiel will be implementing a Scientific workbench that is specialised for GIS research. This will pay special attention to data movement and intergration with existing GIS tools.

Deliverables

  • GIS workbench
  • Middleware for core functionalities
  • Data Flow Facilitator
  • Hierarchical Data Structure
  • Streaming Infrastucture

Timeline

.fx: timeline

Description Start End
Web Presence 25 May 12 June
Initial Feasibilty Demonstration 11 June 29 June
Background Chapter 2 July 29 July
Design Chapter 29 July 29 August
First Implementation 1 July 29 August
Final Implementation 29 August 28 September
Report Outline Complete 28 September 10 October
Report 28 September 31 October
Poster 31 October 3 November
Presentation 11 November 18 November
Jump to Line
Something went wrong with that request. Please try again.