Subscribe to this thread
Home - General / All posts - De-densify a point drawing
171 post(s)
#01-Dec-22 04:23

Hi friends. I have a point drawing with almost a million points. It's a log, from tracking vehicles. I want to do a heatmap, which I do for a webmap. BUT, the txt file is toooo big, 127Mb. So, I import it into M8 and wanted to do like a De-densify... I don't know if that exists.

For example in 3D data, in a point cloud, we can reduce the number of points without losing the form of the shapes... by lowering the number of points used. That is what I want to do with the point drawing in Manifold.

Let me explain a little more, just so you visualize it better. I have a road segment with 12000 points, in the heat map, we may only need to use 3000 for example to show that, that specific road is usually used. Which is the idea of the heatmap.. demonstrate which roads are mainly used by the vehicles tracked.

Anybody with an idea...

Thank you in advance, Tomas.


2,055 post(s)
#01-Dec-22 06:46

Create a new column and fill with random numbers, order by the random numbers and delete the top 9000 records. That should leave 3000 randomly distributed records. If I have though about this correctly!

Aussie Nature Shots

575 post(s)
#01-Dec-22 07:35

Try this: Convert point drawing to las and use PDAL. We once needed certain step between track points (6m). The result was good. There are also other sampling options in PDAL.

pdal translate  locations.las locations_6m.las sample --filters.sample.radius=6

10,027 post(s)
#01-Dec-22 09:22

I would Orthogonalize the points to (say) 0.5m or 1m, then remove duplicates.


10,281 post(s)
#01-Dec-22 13:54

Or Normalize Topology with the desired location tolerance. That should preserve the original locations of the remaining points.

171 post(s)
#13-Dec-22 22:00

Just wanted to say thank you for the recommendations. I used the normalize topology and the random numbers strategies, and worked great. I didn't used the Orthogonalize option, I don't know if it works well with points, anyway will test it soon. And the PDAL, didn't know how to run that.

Thank you friends.

Manifold User Community Use Agreement Copyright (C) 2007-2021 Manifold Software Limited. All rights reserved.