non-uniform terrain mesh simplification using adaptative merge procedures flávio luis de mello...

Non-Uniform Terrain Mesh Simplification Using

Adaptative Merge Procedures Flávio Luis de Mello

Edilberto StraussAntônio OliveiraAline Gesualdi

1. Introduction

2. Mesh Simplification Method

3. Results

4. Conclusion

11

22

Terrain walkthrough plays an important role in the virtual reality

Geographic Information Systems (GIS), Military Mission Planning, Flight Simulation, etc.

Regular grid sampled data is required

33

Relationship between the actual map image resolution and its associated data can easily exceed the capabilities of typical graphics hardware, which makes impossible a real-time interactive application.

data fps real-timeinteraction

44

Relationship between the actual map image resolution and its associated data can easily exceed the capabilities of typical graphics hardware, which makes impossible a real-time interactive application.

data fps real-timeinteraction

At most PII 450MHz

No Hw acceleration

55

Describes an optimized algorithm for building a terrain triangular mesh

Combines an efficient regular grid representation with low cost memory requirements

Bottom-up approach

Based on classic quadtree algorithm (Röttger – Eurograph99)

Digital Elevation Model(2n x 2m)

66

77

h=2

2. Terrain Model

2. Terrain Model

88

Merge Criteria:

#1

#2

2. Terrain Model

99

2. Terrain Model

1010

2. Terrain Model

1111

2. Terrain Model

1212

2. Terrain Model

Sorted

Unsorted

1313

BorderInterior

Gint=0,9143 Gborder=1,3482

hquadtree 4 to 5

1st 1st

2nd 2nd

3rd 3rd

Proposed Röttger Proposed Röttger

iicaseigpG

2. Terrain Model

1414

DEM(1.024 x 1.024)

h=732.768 triangles11.254 triangles#=34,34%#R=16,71%

h=68.192 triangles3.078 triangles#=37,57%#R=9,76%

h=52.048 triangles1.216 triangles#=59,375%#R=4,80%

1515

No cracks, no T-junctions Global treatment

1616

No cracks, no T-junctions Global treatment

1717

DEM(2.048 x 2.048)

=4,18 m h=732.768 triangles15.802 triangles#=48,22%#R=16,03%

h=68.192 triangles5.278 triangles#=64,43%#R=8,84%

h=52.048 triangles1.648 triangles#=80,47%#R=4,51%

1818

DEM(1.024 x 1.024)

=8,125 m h=732.768 triangles26.092 triangles#=79,62%#R=12,47%

h=68.192 triangles7.016 triangles#=85,64%#R=7,69%

h=52.048 triangles1.870 triangles#=91,30%#R=3,73%

1919

The method provides high quality triangulations with thousands of geometric primitives;

The generated mesh is conforming, although it is not well shaped;

Achieved good frame rates on PC platforms, such as 47fps (PII 450MHz, Sthealth II AGP);

Critical future issues include level of detail rendering and efficient paging mechanism, which will allow rendering height fields that do not entirely fit into RAM.

2020