CARTOGRAPHICAL DATA TREATMENT ANALYSISFOR REAL TIME THREE-DIMENSIONAL

TERRAIN VISUALIZATION

Flávio Luis de Mello & Luiz Felipe Coutinho Ferreira da Silva

Divisão de Tecnologia da Informação - DTI & Seção de Engenharia Cartográfica - SE6Centro Tecnológico do Exército - CTEx & Instituto Militar de Engenharia - IME

11

Summary

1. Introduction

2. Real Time Systems

3. The Terrain Mesh

4. The Cone Vision

5. The Triangulation Construction Study Case

6. Results

7. Conclusion

Introduction

• Alternative solution for the triangulation problem of altimetric data of cartographic features

• Main goal: allow data management obtain real time 3D terrain visualization from 2D topographic charts

• Application is supposed to be executed on low performance computers embedded on military tanks, vehicles and soldiers combat equipment

• The methodology is constrained by this obsolete hardware (upper bound: Pentium 300 MHz) 22

33

Real Time Systems

• Application metric: scene refresh rate

ExecutionTime

industrial and electrical plants,intensive therapy units

elastic timeresponse

maritime traffic control,cosmic probes navigation snapsho

tresponse definition is not associated to the elapsed

time, but to an opportune problem treatment

• Real time animation: over 12 fps

44

The Terrain Mesh

• Impossible to know exactly the heights of all points on a surface

• Attribute the same height to all neighboring points near a sample height

• Each point is connected using straight line segments in order to compose a triangle set, and each triangle vertex is raised until its corresponding height

• Each triangle vertex corresponds to a point measure on the object surface

44

The Terrain Mesh

• Impossible to know exactly the heights of all points on a surface

• Attribute the same height to all neighboring points near a sample height

• Each point is connected using straight line segments in order to compose a triangle set, and each triangle vertex is raised until its corresponding height

• Each triangle vertex corresponds to a point measure on the object surface

Delaunay triangulation

55

The Cone Vision

• Scene vision composed by several layers

• Complex scenes

• If it is possible to determine the invisible geometries before being drawn, it would be possible simply to ignore them, thus preventing the unnecessary data processing

66

Spatial Indexing

77

The Triangulation Construction Study Case

Piraquara Extract (MI-2842-4-SE), 1:25,000

2,478 points

88

The Triangulation Construction Study Case

2,478 points 5,173 triangles

Piraquara Extract (MI-2842-4-SE), 1:25,000

88

The Triangulation Construction Study Case

2,478 points 5,173 triangles

Piraquara Extract (MI-2842-4-SE), 1:25,000

99

Results

• The time spent by the graphical device to draw the scene over the monitor screen is inversely influenced by the quantity of geometrical primitives

Frustum culling algorithmDouglas-Peucker generalization

contributes for reducing the number of geometries necessary for the exhibition

• Altimetry data represents about 76% of all the map data (average)

1010

Results

61.181 points 135.504 triangles

Piraquara (MI-2842-4-SE), 1:25,000

Test = <processor, video board, grid>

1 GHz233 MHz

GeForce2 64MB PCIGeForce2 64MB AGPS3Virge 1MB PCI

1x1,2x2,4x4,8x8,16x16,32x32,64x64,128x128

1111

Results

1212

Results

1313

Conclusion

• It was possible to improve the application performance to higher values than the lower limits defined by classic animation literature

• The results analysis suggests that it is possible to create a real time 3D terrain visualization application based on the current available cartographic data and obsolete hardware

• Future work includes: remaining data layers, cartographic symbology, texture rendering, database transaction.

1414

Questions

_________ ________________ ___ ____________ ______ ____________

___ ______________ _______________

____________ ___________ ___