[Therion] Turning LIDAR point clouds into cave maps

[Martin Budaj]

On Wed, Nov 29, 2023 at 4:47 PM Bill Gee <bgee at campercaver.net> wrote:

> Looking at the sample files, it looks like he extracts some images from
> the point cloud.  One of them is an overhead view and others are cross
> sections and profiles.  He then uses these JPG images as the drawing
> background to produce a traditional cave map.  The LiDAR scan
> essentially replaces the in-cave sketching.  I suspect there is some
> custom software to extract the images from the point cloud.

Indeed, the scans (combined with videogrammetry as in iPhone or iPad)
preserve a lot of details and can be used to draw 2D maps. However,
they can't completely replace the sketches or notes, as there is much
information that can't be captured in the scans (air droughts, lake
bottom etc.)

> I remember some years ago hearing about some mapping projects where the
> survey stations were marked with small balls in a specific color.  The
> processing software recognized that color and provided a way to connect
> those points to known x,y,z locations - a centerline.  It was
> fantastically expensive in both money and computer resources.
>
> In a way, we do sort of the same thing with Therion.  The centerline
> data is processed to produce a set of x,y,z coordinates.  The survey
> stations are flagged in that data set.  Then when the sketch is drawn
> out, we insert points of type "survey station" which are also flagged.
> The sketch can be considered as a sort of two dimensional point cloud.
> Therion knows how to match up the survey station points in the sketch
> with the survey stations in the x,y,z coordinate set.  From there it can
> morph everything around.

This is exactly the plan how to implement it in Therion, when time permits:

1) you scan a fragment of a cave (a 3D scrap) with some survey
stations visually marked
2) you assign station names as used in Therion centreline to local 3D
coordinates in the 3D scrap
3) Therion warps the 3D scrap and aligns it with the centreline
4) ideally, Therion smoothly joins adjacent 3D scraps

The result would be a 3D model which could combine the walls captured
from scans with those generated from the 2D scraps and LRUD data. This
way you could progressively improve the 3D model as your LiDAR
scanning progresses.

There is a preliminary issue for this here:
https://github.com/therion/therion/issues/475

Martin

P.S. Here is a 50-metres dig scanned by an Apple LiDAR in around 30
minutes: https://photos.app.goo.gl/Cvs8mej2RCqpGoj9A