Hello,
I apologize if this is not the right place to ask this question. I'm using CGAL to compute a constrained delaunay triangulation over the XY plane for the purpose of turning a point cloud into a 2.5D mesh. Here are two screenshots to illustrate: Point cloud: https://cloud.githubusercontent.com/assets/1951843/24327134/0bc19458-1197-11e7-96ae-4df1c1bd99a1.png 2.5D mesh calculated from the point cloud: https://cloud.githubusercontent.com/assets/1951843/24327136/1448b174-1197-11e7-8678-e2e2c0bb5292.png My goal at this moment is to smooth the resulting mesh as to remove some of the "spikes" that appear after the 2.5D mesh is computed. I've circled a few examples in this screenshot: https://cloud.githubusercontent.com/assets/1951843/24327146/6203f9dc-1197-11e7-9c19-11e41d4e6d2c.png I'm fairly new to CGAL and meshing algorithms, so I was hoping some more experienced users could point me in the right direction? Thank you! -- You are currently subscribed to cgal-discuss. To unsubscribe or access the archives, go to https://sympa.inria.fr/sympa/info/cgal-discuss |
Hello,
Don't worry, this is the right place to ask this question. From what I see in your pictures, I would use the connectivity information of the triangulation to remove these artifacts. More specifically, circulate over the incident vertices: http://doc.cgal.org/latest/Triangulation_2/classCGAL_1_1Triangulation__2.html#aba34806b0052ed4d1b7c07f2626b1a38 (For an example on how to use the circulator, see here: http://doc.cgal.org/latest/Triangulation_2/index.html#title9 ) For each vertex, you can check if the height between this vertex and its incident vertices is greater than a threshold and remove it if this is the case. If you don't want to remove points, I think replacing the height of the vertex by the median height of its incident vertices can be a good solution too. Let me know if I'm not clear or if you need more help. Best, -- Simon Giraudot, PhD R&D Engineer GeometryFactory - http://geometryfactory.com/ Le 26/03/2017 à 01:23, Piero Toffanin a écrit : > Hello, > > I apologize if this is not the right place to ask this question. > > I'm using CGAL to compute a constrained delaunay triangulation over > the XY plane for the purpose of turning a point cloud into a 2.5D > mesh. Here are two screenshots to illustrate: > > Point cloud: > https://cloud.githubusercontent.com/assets/1951843/24327134/0bc19458-1197-11e7-96ae-4df1c1bd99a1.png > > 2.5D mesh calculated from the point cloud: > https://cloud.githubusercontent.com/assets/1951843/24327136/1448b174-1197-11e7-8678-e2e2c0bb5292.png > > My goal at this moment is to smooth the resulting mesh as to remove > some of the "spikes" that appear after the 2.5D mesh is computed. I've > circled a few examples in this screenshot: > https://cloud.githubusercontent.com/assets/1951843/24327146/6203f9dc-1197-11e7-9c19-11e41d4e6d2c.png > > I'm fairly new to CGAL and meshing algorithms, so I was hoping some > more experienced users could point me in the right direction? > > Thank you! > > -- You are currently subscribed to cgal-discuss. To unsubscribe or access the archives, go to https://sympa.inria.fr/sympa/info/cgal-discuss |
In reply to this post by Piero Toffanin
On 03/26/2017 01:23 AM, Piero Toffanin wrote:
> Hello, > > I apologize if this is not the right place to ask this question. > > I'm using CGAL to compute a constrained delaunay triangulation over the > XY plane for the purpose of turning a point cloud into a 2.5D mesh. Here > are two screenshots to illustrate: > > Point cloud: > https://cloud.githubusercontent.com/assets/1951843/24327134/0bc19458-1197-11e7-96ae-4df1c1bd99a1.png > > > 2.5D mesh calculated from the point cloud: > https://cloud.githubusercontent.com/assets/1951843/24327136/1448b174-1197-11e7-8678-e2e2c0bb5292.png > > > My goal at this moment is to smooth the resulting mesh as to remove some > of the "spikes" that appear after the 2.5D mesh is computed. I've > circled a few examples in this screenshot: > https://cloud.githubusercontent.com/assets/1951843/24327146/6203f9dc-1197-11e7-9c19-11e41d4e6d2c.png > Just to be sure: the noise is already in the input point set, right? Sebastien. > > I'm fairly new to CGAL and meshing algorithms, so I was hoping some more > experienced users could point me in the right direction? > > Thank you! > > -- You are currently subscribed to cgal-discuss. To unsubscribe or access the archives, go to https://sympa.inria.fr/sympa/info/cgal-discuss |
In reply to this post by Simon Giraudot-2
Simon,
thank you for the explanation! It makes sense. Like you suggested, I would probably choose to use the median height rather than eliminating the vertices. I will attempt to write your proposed solution and I'll come back if I have questions. On 03/27/2017 01:38 AM, Simon Giraudot wrote: > Hello, > > Don't worry, this is the right place to ask this question. > > From what I see in your pictures, I would use the connectivity > information of the triangulation to remove these artifacts. More > specifically, circulate over the incident vertices: > http://doc.cgal.org/latest/Triangulation_2/classCGAL_1_1Triangulation__2.html#aba34806b0052ed4d1b7c07f2626b1a38 > > (For an example on how to use the circulator, see here: > http://doc.cgal.org/latest/Triangulation_2/index.html#title9 ) > > For each vertex, you can check if the height between this vertex and > its incident vertices is greater than a threshold and remove it if > this is the case. If you don't want to remove points, I think > replacing the height of the vertex by the median height of its > incident vertices can be a good solution too. > > Let me know if I'm not clear or if you need more help. > > Best, > -- You are currently subscribed to cgal-discuss. To unsubscribe or access the archives, go to https://sympa.inria.fr/sympa/info/cgal-discuss |
In reply to this post by Sebastien Loriot (GeometryFactory)
Yes, the noise is already in the input point set. On 03/27/2017 03:57 AM, Sebastien
Loriot (GeometryFactory) wrote:
On 03/26/2017 01:23 AM, Piero Toffanin wrote: --
Piero Toffanin
masseranolabs.com |
Free forum by Nabble | Edit this page |