Box Volume Test

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

Box Volume Test

Post by Matt Markworth » Sun Mar 29, 2015 11:03 am

NTS,

The goal of this test was to compare the calculated volume of a box using two methods. The first method is the standard volume measurement of length x width x height. The second method consists of the creation and measurement of a 3D model using a combination of three software programs: 123D Catch, Meshmixer, and Meshlab.

Using a box provides the advantage of getting a more accurate standard measurement compared to an irregularly shaped section of a tree.

Method #1
- Took multiple measurements of each side and averaged them. This is why the measurements are in tenths of millimeters.
- 32.81 cm x 23.32 cm x 23.18 cm = 17736 cm3

Method #2
- Took photos and used 123D Catch to create a mesh
- Cleaned up the mesh in Meshmixer
- Scaled the mesh in Meshlab
- Used Meshlab to calculate a volume of 17799 cm3
Meshlab result
Meshlab result
Here are two videos showing the box. The first one shows the texture of the surfaces (images from the photographs) and the second one just shows the wireframes.

I drew an Egyptian alphabet on the sides as a way to help the software stitch the photos and also just to make the box more interesting.

https://www.youtube.com/watch?v=yUFa4TZmgnA
https://www.youtube.com/watch?v=yUFa4TZmgnA

https://www.youtube.com/watch?v=wYIXrzK-5xg
https://www.youtube.com/watch?v=wYIXrzK-5xg

Matt

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

Re: Box Volume Test

Post by Don » Sun Mar 29, 2015 1:34 pm

Matt-
My interpretation of the results is that Method #2 was the more accurate measure of volume, assuming that it took a large number of measurements that would have included the imperfect planes of an old box?
-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: Box Volume Test

Post by Matt Markworth » Sun Mar 29, 2015 8:44 pm

Don,

Because this box isn't a perfect geometric shape, theoretically Method #2 should give the more accurate volume measurement. However, there are at least two main sources of error that can affect the outcome.

1) The mesh is a close representation of the box, however the surfaces of the mesh are a little more bumpy than the box.

2) There is some error in scaling the mesh, although it can be reduced with care.

Meshlab will give an exact volume of the mesh, however the volume measurement is in purely arbitrary units. This arbitrary volume measurement can be done before any measurements of the actual object have been made. At least one measurement must be made on the object (preferably more) and then those same reference points measured within Meshlab with the measuring tool.

In this example, I measured the length of the box as 32.8 cm. I used the measuring tool in Meshlab and measured the length of the mesh as 32 arbitrary units. 32.8 divided by 32 gives a scaling factor of 1.025. This scaling factor is entered into the Transform Scale function. The mesh is now scaled in cm and a volume measurement can be done in cm. The biggest source for error in this process is if the reference measurement on the mesh isn't done carefully.

I was especially pleased to complete this test because creating the mesh consisted of using completely different software compared to my last test: 123D Catch/Meshmixer vs. Visual SFM/Meshlab. Having various ways to create a mesh should prove useful for comparison purposes in the future.

Matt

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

Re: Box Volume Test

Post by Don » Sun Mar 29, 2015 10:28 pm

Matt-
Extending the discusssion, the take-home for me is if we wish to use SfM and it's variants for highly accurate volume measurements of irregular shapes, we need a reference object in the image to base the volume calculations on? As an example, a cylinder of known dimension (say a soup can) adjacent to the bole being measured?
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: Box Volume Test

Post by Matt Markworth » Mon Mar 30, 2015 5:44 pm

Don,

Using a reference object or measuring the straight line distance between two easily distinguishable characteristics on the tree will work. Having a couple reference objects would be ideal for getting a very accurate scaling ratio. Separate ratios could be calculated and then averaged.

Also, I uploaded the mesh to Sketchfab if you'd like to peruse the surfaces: https://sketchfab.com/models/3785bcf450 ... 84c6bd4d06

Matt

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

Re: Box Volume Test

Post by Don » Mon Mar 30, 2015 6:39 pm

Matt-
That was cool!
I was running this on my Mac Air 11", and after experimenting with single and dual finger slide controls, I was able to manipulate images in a 3D space, circle around it, go through it, etc. The box was cool, the sweet gum segment had excellent bark surface, but I liked the silver maple base the best. I get the sense that there's no one tool to do everything, just a toolbox with separate specialized tools to work on the images we'd bring forth.
What tool would be useful in measuring dimensions, and by extension, volume?
-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: Box Volume Test

Post by Matt Markworth » Mon Mar 30, 2015 7:17 pm

Don,

Thanks. I like your toolbox analogy.

Of the software that I've tested, here's how they fit into the toolbox:

Turning photos into meshes: VisualSFM, 123D Catch
Manipulating meshes: Meshmixer, MeshLab, Blender
Measuring meshes (including volume): MeshLab
Presentation: Videos from 123D Catch, Sketchfab

Matt

Post Reply

Return to “Measurement and Dendromorphometry”