Re: Comparing two volume measurements

General discussions of measurement techniques and the results of testing of techniques and equipment.

Moderators: edfrank, dbhguru

Post Reply
User avatar
Matt Markworth
Posts: 1302
Joined: Fri Nov 23, 2012 8:41 pm

Comparing two volume measurements

Post by Matt Markworth » Mon Mar 23, 2015 9:49 pm

NTS,

This evening I ran a test comparing two volume measurement techniques. My goal was to see how close the measurements would be and to think about some of the variables associated with each technique. I've listed some of those variables at the end. The object being measured was a 3' section of a sweetgum trunk. A yardstick was thumbtacked to the trunk for scale.

Technique #1
- Measure the circumference at the top and bottom of the yardstick, then derive radius
- Plug the #'s into the CalculatorSoup.com frustum calculator
- Volume calculated as 7.286 ft³
Sweetgum
Sweetgum
Calculator Soup Result
Calculator Soup Result
Technique #2
- Take photos around the tree
- Use Visual SFM to stitch the photos
- Use Meshlab to create a mesh
- Use the yardstick impression in the mesh to create a scaling ratio
- Find volume in Meshlab
- Volume calculated as 7.242 ft³
Visual SFM Sparse Reconstruction
Visual SFM Sparse Reconstruction
Meshlab View
Meshlab View
meshlab - cleaned up.JPG (37.21 KiB) Viewed 849 times
Meshlab Result
Meshlab Result
Here are some variables affecting the results:
- The yardstick wasn't completely vertical, it was at a slight angle
- Slight variances with tape wrap accuracy
- The shape is similar to a frustum, but of course, not a perfect frustum
- The tape wrap includes furrows in the bark, inflating Technique #1
- This tree has an indent on one side, caused by root girdling, inflating Technique #1
- The slight imperfection of cutting off the top and bottom of the mesh right at the edge of the tape
- The top and bottom of my mesh were a bit "puffy," I need to figure out how to make them flat, thus inflating Technique #2

I want to test on a larger scale in the future, but thought it would be good to start with some smaller tests and work from there. You may notice some letters/numbers on the tape. I haven't tested it too much, but I think this helps Visual SFM find similarities across the photos making it easier for the software to stitch them together.

Matt

User avatar
dbhguru
Posts: 4464
Joined: Mon Mar 08, 2010 9:34 pm

Re: Comparing two volume measurements

Post by dbhguru » Tue Mar 24, 2015 11:27 am

Matt,

Great job! You're out there on the cutting edge. I downloaded MeshLab for the MAC about a year ago, but never got around to really exploring it as a tool for volume modeling. I just downloaded a demonstration copy of Photoscan Professional from Agisoft. The advantage of this software is presumably that it's easier to use. However, it isn't free and I haven't seen its price tag. They don't tell you that up front. However, the manual looks very well organized. I'll report on progress as familiarization proceeds.

For the more conventional trunk modeling, remember that there are two other frustum models that may be applicable, i.e. the paraboloid and the neiloid. For the section near the base of a large tree, the neiloid form is likely to fit best, especially if the roots fare a lot. However, If you model using a neiloid form you might need a different formula for a frustum. It is messy.

If Au = area of upper base, Ab= area of lower base, and h = frustum height, the frustum's volume is given by
Screen shot 2015-03-24 at 12.15.41 PM.png
We can program in messy formulas like this into an iPhone spreadsheet or another software app.

As Don Bertolette has frequently reminded me, we need to bring this kind of technology to Cadre members. It won't be for everyone, but a few need to be able to calculate trunk volumes this way.

Bob
Robert T. Leverett
Co-founder, Native Native Tree Society
Co-founder and President
Friends of Mohawk Trail State Forest
Co-founder, National Cadre

User avatar
Matt Markworth
Posts: 1302
Joined: Fri Nov 23, 2012 8:41 pm

Re: Comparing two volume measurements

Post by Matt Markworth » Tue Mar 24, 2015 7:29 pm

Thanks Bob, I'll read up on some of the shapes you mentioned. It would be fun to do tests on those shapes comparing traditional volume measurement versus a 3D model technique.

It's pretty easy to create a model in Meshlab, but the learning curve gets steeper when trying to fine tune things.

Here are some steps for those interested in giving Visual SFM/Meshlab a try:

- Download Visual SFM
- File
- Open + Multiple Images
- Click 4 arrowed icon (Compute Missing Matches)
- Click double arrowed icon (Compute 3D Reconstruction)
- CMVS icon, give it a file name, click save
- Tab key to switch between sparse and dense cloud
- Download Meshlab
- File, Open Project
- 00, bundle.rd.out, then list
- Icon that looks like layers (Show Layer Dialog)
- Right click and choose delete current mesh
- File, Import Mesh, Models, import .ply file
- To clean up Mesh use Select Vertexes icon, then the delete icon (a triangle with an X over it)
- File, export mesh as, give a new name, save, ok
- Filters, Point Set, Surface Reconstruction: Poisson, Apply
- Select, non-manifold edges, delete them
- Use Fill Holes icon
- Once it's "watertight", the volume can be found after being scaled
- take real measurement of 2 reference points divided by the measurement of those 2 reference points on the mesh
- this gives a ratio that is used as the scaling factor
- filters
- normals, curvatures
- transform scale
- put ratio in x axis
- checkmark uniform scaling
- checkmark freeze matrix to make scaling permanent
- Filters, Quality Measures and Computations, Compute Geometric Measures

I need to learn more about fine tuning things. Here is the "watertight" sweetgum trunk model that you can move around and inspect how it looks. You'll notice the ends are puffy and one end has a small cavity for some reason.

Here is the link for the model uploaded to Sketchfab:

https://sketchfab.com/models/25a81aff3b ... 375cafdab3

Matt

User avatar
Matt Markworth
Posts: 1302
Joined: Fri Nov 23, 2012 8:41 pm

Re: Comparing two volume measurements

Post by Matt Markworth » Fri Mar 27, 2015 6:01 pm

All,

Just for fun, here's a video of the 3' section of sweetgum trunk using 123D Catch:

https://www.youtube.com/watch?v=zf86ORTZ5OM
https://www.youtube.com/watch?v=zf86ORTZ5OM

Matt

User avatar
Don
Posts: 1560
Joined: Tue Mar 09, 2010 12:42 am

Re: Re: Comparing two volume measurements

Post by Don » Fri Mar 27, 2015 11:11 pm

Mark-
Impressive! I really like the rotational aspect, especially with a measure of transparency. Looks like it takes about 16 images or so? Is there a certain amount of overlap needed to get the 'fluency'?
-Don
Don Bertolette - President/Moderator, WNTS BBS
Restoration Forester (Retired)
Science Center
Grand Canyon National Park

BJCP Apprentice Beer Judge

View my Alaska Big Tree List Webpage at:
http://www.akbigtreelist.org

User avatar
Matt Markworth
Posts: 1302
Joined: Fri Nov 23, 2012 8:41 pm

Re: Re: Comparing two volume measurements

Post by Matt Markworth » Sat Mar 28, 2015 6:44 am

Don wrote:Mark-
Impressive! I really like the rotational aspect, especially with a measure of transparency. Looks like it takes about 16 images or so? Is there a certain amount of overlap needed to get the 'fluency'?
-Don
Don,

Thanks. I really like how the software allows you to set different levels of transparency. You can select wireframes only, wireframes and texture, or simply just texture. There is also a sliding bar to increase or decrease each of those settings. For example, you can use the sliding bar to decrease texture, which reduces the impact of the photograph on the image and increases the impact of the wireframes. There is also an option called 2-Sided Surfaces. Unchecking that will eliminate the transparency so that the side of the tree in the background can't be seen.

Yes, 16 photos can create a good mesh. In order to make sure I had a good mesh for this one, I took ~16 from eye level and also ~16 from a couple feet below that. I'm not sure of the minimum amount of overlap. I think it depends on whether the object has a lot of distinguishable features (e.g. American sycamore bark) or if the object is very similar across it's surface. Getting both horizontal and vertical overlap with the photos helps, and you can never have too much overlap.

Here's another video where you will be transported through the tree:

https://www.youtube.com/watch?v=-ycMjxuKjsU
https://www.youtube.com/watch?v=-ycMjxuKjsU

Matt

Joe

Re: Re: Comparing two volume measurements

Post by Joe » Sat Mar 28, 2015 8:14 am

hmmm.... here's a crazy idea- but--- some of you are arborists- and I'm sure you often drop trees for clients- perhaps you could try out these measurement software tools on the standing tree, then drop the tree, then cut up the tree into small pieces and measure the volumes in large vats of water..... or weigh the wood if you have a way to do that...
Joe

User avatar
Don
Posts: 1560
Joined: Tue Mar 09, 2010 12:42 am

Re: Comparing two volume measurements

Post by Don » Sun Mar 29, 2015 12:36 am

Matt-
It's late, but I have to say, my first thought after viewing this most recent video was
'Beam me up, Scotty"?
A little 'Transporter" music and I'd have been gone...

So the slide mechanism that adjusts transparency is what makes the TIN appear to be bark, almost a surrogate for bark thickness...

Which is easier, measuring volume of bole segments by displacement, or by computing them through SfM/123D Catch/MeshLab softwares?
All very cool !
-Don
Don Bertolette - President/Moderator, WNTS BBS
Restoration Forester (Retired)
Science Center
Grand Canyon National Park

BJCP Apprentice Beer Judge

View my Alaska Big Tree List Webpage at:
http://www.akbigtreelist.org

Post Reply

Return to “Measurement and Dendromorphometry”