12

Is there any package/software which can do Meshing of Point Clouds in real time?

What is the data structure used to represent 3D Point Clouds ?

iceman
  • 4,211
  • 13
  • 65
  • 92

5 Answers5

11

MeshLab is a great one for meshing point clouds.

Look under Filters > Re-meshing, simplification and reconstruction. Ball Pivoting Surface Reconstruction and Poisson Reconstruction are good ones.

Here you can find a tutorial on Meshing Point Clouds using MeshLab.

As for representing your point cloud the easiest format to use is the XYZ format:

X1 Y1 Z1
X2 Y2 Z2
etc.

Another good format that is relatively easy to use is Object (OBJ) model format.

worbel
  • 6,509
  • 13
  • 53
  • 63
  • 1
    I used it to reconstruct alpha shapes, but the `asc` format didn't worked for me. Thanks for mentioning the `xyz` format here, which did the trick. – math Mar 04 '14 at 15:09
4

MeshLab http://meshlab.sourceforge.net/

Laureano
  • 247
  • 2
  • 4
  • can it do that real-time..as the data is acquired from the lidar – iceman Apr 02 '10 at 17:35
  • 1
    @iceman For typical applications you are probably going to take several scans and then manually register them off-line to make one big point cloud before you think about meshing them. A single scan is limited by occlusions in the line of sight. – ahcox Apr 15 '12 at 21:32
3

Marc Levoy's website contains links to packages used to create several famous 3D models, but they are not real time. In particular, VripPack and ZipPack. He has some papers on doing it in real time too, I think, but they lack packaged implementations.

Victor Liu
  • 3,545
  • 2
  • 24
  • 37
0

3DRACS is the best option. try it Its easy and fast. http://real3d.pk/3dracs.html

maxpayne
  • 1,111
  • 2
  • 21
  • 41
0

Geomagic can do it very well http://www.geomagic.com, but not in real time and it is very expensive. I think a lot of users of lidar data are moving away from meshing and are focused improving the methods of viewing point-clouds- extracting surface normals to allow lighting and scaling up the point sizes intelligently to fill holes.

Lucas Walter
  • 942
  • 3
  • 10
  • 23