HoughNormals plugin algorithm
Posted: Thu Jul 23, 2026 10:45 am
Hello,
(First, thank you for all your work, apologies if this seems like nitpicking but it may be relevant to my research)
I am trying to figure out exactly which algorithm has been implemented in the HoughNormals plugin so that I can make informed decisions about parameters.
The wiki page for the HoughNormals plugin says that it uses the algorithm presented in "Normal Estimation in Unstructured Point Clouds with Hough transform by A. Boulch and R. Marlet", this paper doesn't exist. The title is, however, the same as the summary used at the top of the github page linked, which references the paper "Fast and Robust Normal Estimation for Point Clouds with Sharp Features by Alexandre Boulch and Renaud Marlet", in which three algorithms were presented. Further confusing matters is that the figure on the wiki is from a later paper "Deep Learning for Robust Normal Estimation in Unstructured Point Clouds by Alexandre Boulch and Renaud Marlet" which presents a newer algorithm, which has its own, separate github page.
Does anyone know which of the 4 algorithms between these two papers is actually implemented in the plugin? My best guess is that it is RRHT_c from the earlier paper given the existence of the tolerance angle parameter, however the plugin features a "neighbourhood size for density estimation" parameter, which appears as a parameter in the later paper.
Thank you for your time,
Rufus
(First, thank you for all your work, apologies if this seems like nitpicking but it may be relevant to my research)
I am trying to figure out exactly which algorithm has been implemented in the HoughNormals plugin so that I can make informed decisions about parameters.
The wiki page for the HoughNormals plugin says that it uses the algorithm presented in "Normal Estimation in Unstructured Point Clouds with Hough transform by A. Boulch and R. Marlet", this paper doesn't exist. The title is, however, the same as the summary used at the top of the github page linked, which references the paper "Fast and Robust Normal Estimation for Point Clouds with Sharp Features by Alexandre Boulch and Renaud Marlet", in which three algorithms were presented. Further confusing matters is that the figure on the wiki is from a later paper "Deep Learning for Robust Normal Estimation in Unstructured Point Clouds by Alexandre Boulch and Renaud Marlet" which presents a newer algorithm, which has its own, separate github page.
Does anyone know which of the 4 algorithms between these two papers is actually implemented in the plugin? My best guess is that it is RRHT_c from the earlier paper given the existence of the tolerance angle parameter, however the plugin features a "neighbourhood size for density estimation" parameter, which appears as a parameter in the later paper.
Thank you for your time,
Rufus