remove noise and outlier

[s-wtnb]

What setting of PoissonRecon or SurfaceTrimmer should be done to remove small noise or outliers as shown in the figure below?

[mkazhdan]

For the reconstruction code, you can try to set the --pointWeight dlag to something less than the default value of 4. Maybe even zero if the input is noisy.

If you run the reconstruction with the --density flag enabled, you can try to set the --aRatio value in the trimming code to something higher than the default value of 0.001.

-- Misha

On June 25, 2018 4:30:06 AM EDT, s-wtnb notifications@github.com wrote:

What setting of PoissonRecon or SurfaceTrimmer should be done to remove small noise or outliers as shown in the figure below?

-- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/mkazhdan/PoissonRecon/issues/70

[s-wtnb]

thx! Although it did not disappear by changing aRatio, the noise disappeared by setting pointWeight to 0.

By the way, what else should I do to increase the number of points using the coordinates of the original point group as much as possible?

[mkazhdan]

You could also try to set the value for the --samplesPerNode parameter higher.

On June 25, 2018 8:51:20 PM EDT, s-wtnb notifications@github.com wrote:

thx! Although it did not disappear by changing aRatio, the noise disappeared by setting pointWeight to 0.

By the way, what else should I do to increase the number of points using the coordinates of the original point group as much as possible?

-- You are receiving this because you commented. Reply to this email directly or view it on GitHub: https://github.com/mkazhdan/PoissonRecon/issues/70#issuecomment-400141368

[s-wtnb]

thx. Are there settings that uses raw data as much as possible(= not smoothing)?

[mkazhdan]

Sure.

You can try setting the depth to a high value and setting the number of samples per node to zero. (You can also try setting the point weight to a large value, though that may result in over-fitting.)

-- Misha

From: s-wtnb notifications@github.com Sent: Friday, June 29, 2018 12:14 AM To: mkazhdan/PoissonRecon PoissonRecon@noreply.github.com Cc: mkazhdan misha@cs.jhu.edu; Comment comment@noreply.github.com Subject: Re: [mkazhdan/PoissonRecon] remove noise and outlier (#70)

thx. Are there settings that uses raw data as much as possible(= not smoothing)?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/mkazhdan/PoissonRecon/issues/70#issuecomment-401242463 , or mute the thread https://github.com/notifications/unsubscribe-auth/AKXc5XJkHO3cHba2d3dyWoNUF8ll_cnMks5uBamFgaJpZM4U1uEO .