Photo Measurements

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

Moderators: edfrank, dbhguru, edfrank, dbhguru

#1)  Photo Measurements

Postby dbhguru » Mon Jan 07, 2013 5:35 pm

NTS,

   My primary camera is a Canon Powershot SX260 HS. I've been testing it as a means of measuring diameter at a distance. So far, it has worked. Here is an image taken outdoors. You see two rulers in the image. One ruler is 12 inches and the other 18. I used the 18-inch ruler as a reference object. The 12-inch ruler is the target object, i.e. the object to be measured through photographic analysis.

               
                       
PhotoMeasurement-5.jpg
                                       
               


   I first pull the image into Excel. Using shape objects, I then mask the reference and and target objects with line objects and set their colors to yellow so they will show up clearly. By going to the format option and size sub-option, I can take the Excel size of each object. With my laser rangefinder, I take the distance of each object. Using the formula shown, I then calculate the actual size of the target object using the Excel size and the other quantities. So far, the system has worked well and is independent of the how much of the telephoto capability I use. I'll show the results of more tests in the coming weeks. Note in the image, I was sloppy in orienting the reference and target objects and still came within 0.4 inches of the actual length of the target.

   This method works for circular/cylindrical objects at varying distances. You have to have the distance to each target, the actual size of the reference object and its distance. If you have those quantities, you could measure diameters and corresponding girths of all the trees in your photograph. At least, I can with my camera. What you can do for the trunk of a tree, you can do for a limb. So basically, if you identify the points on a trunk and its limbs that you want to measure and record their distances, vertical angles, and azimuths, so long as you have your reference object and its distance is n the photo, you can model what you see in the photograph. This method provides us with an alternative to using a reticle focused monocular, which can be time consuming, and in addition to the rangefinder and clinometer requires a tripod to steady the monocular.

    With Visual Basic for Applications, the process could be partially automated. It would require a rigidly followed measuring protocol, but isn't beyond the realm of possibility. Michael Taylor is better at this kind of sophisticated graphics programming than I am, but Michael has his hands full. If he doesn't take it on as a project, I may give it a try.

     But we don't have to wait on an automated solution. Members can use the method described above to some advantage.

Bob
Robert T. Leverett
Co-founder and Executive Director
Native Native Tree Society
Co-founder and President
Friends of Mohawk Trail State Forest

For this message the author dbhguru has received Likes - 3:
edfrank, ElijahW, jamesrobertsmith
User avatar
dbhguru
 
Posts: 3048
Joined: Mon Mar 08, 2010 10:34 pm
Location: Florence, Massachusetts
Has Liked: 0 times
Has Been Liked: 733 times
Print view this post

#2)  Re: Photo Measurements

Postby dbhguru » Wed Jan 09, 2013 10:22 am

NTS,

    Below is another test of a simple photographic measurement process. The tree is the Reems Creek Poplar. I tried to identify a reasonable mid-slope position and locate a point 4.5 feet above on the trunk. I used 71 inches as the arm spread of the gent in the image. Why 71? Well, I'm pretty sure that the average arm span for men during the period when the photo was taken was between 68 and 74 inches. I went midway between. In addition, the reference object (stretched arms) lies in a vertical plane that is closer to the camera than the vertical plane through the center of the tree. So the from the camera to the reference and target objects are different. I treated them as the same in the formula. In addition, I don't know where on the trunk the measurement of  28.7 feet was taken. I have no way of knowing if these sources of error complement or cancel. The fact that the photographic result comes within 0.5 feet of the taped result has to be pure luck. But we can still get into the ball park even when we don't know the values of the constituent variables.    

               
                       
Screen shot 2013-01-08 at 2.55.34 PM.png
                                       
               


Bob
Robert T. Leverett
Co-founder and Executive Director
Native Native Tree Society
Co-founder and President
Friends of Mohawk Trail State Forest

For this message the author dbhguru has received Likes :
bbeduhn
User avatar
dbhguru
 
Posts: 3048
Joined: Mon Mar 08, 2010 10:34 pm
Location: Florence, Massachusetts
Has Liked: 0 times
Has Been Liked: 733 times
Print view this post

#3)  Re: Photo Measurements

Postby edfrank » Thu Jan 10, 2013 12:33 pm

Bob,

I have used similar techniques to estimate the diameter of trees in photos using the pixel counts in my graphics program (Thumbs Plus).  These photos have square pixels so the vertical and horizontal pixel distances should be the same.  I did not get quite the same numbers as you did for the marked lines, but perhaps this is simply a graphing error as the amount was small.  Using your horizontal line as 71 inches.  There are 2.35 pixels for every inch.  That would make he man 6 foot 6" tall, and the girth at the line 27.7 feet.   If the man was shorter than 6' 6", then the diameter would be smaller.    

The process using photo pixels is easy.  Most graphics programs will give you a pixel count.  If the measured distance is exactly horizontal or vertical the pixels can be directly read off the count indicator.  In this case the horizontal line for arm spread was 233 - 66 = 167 pixels = 2.35 pixels per inch  assuming an arm spread of 71 inches.  The girth was 282 - 32 pixels = 250 pixels  = diameter 106 inches = girth 27.7 feet.  The mans height was 184 pixels or 6' 6".  If the measured distance is not directly horizontal then you can use the Pythagorean Theorem:

square root[(horizontal pixel difference)2 + (vertical pixel difference)2] = distance in pixels.

It is pretty straight forward to implement.  I am not sure exactly how you made your measurements, but I don't see that this method would be any less accurate than any other method as the major potential errors are in guessing the size of objects used for scale, and choosing the endpoints of your measurement sections.

Edward Frank

.
"I love science and it pains me to think that so many are terrified of the subject or feel that choosing science means you cannot also choose compassion, or the arts, or be awe by nature. Science is not meant to cure us of mystery, but to reinvent and revigorate it." by Robert M. Sapolsky
User avatar
edfrank
Site Admin
 
Posts: 4100
{ IMAGES }: 0
Joined: Sun Mar 07, 2010 6:46 pm
Location: Reynoldsville, Pennsylvania, USA
Has Liked: 822 times
Has Been Liked: 600 times
Blog: View Blog (3)
Print view this post

#4)  Re: Photo Measurements

Postby fooman » Thu Jan 10, 2013 4:11 pm

All,

I would highly recommend  imagej as a digital image manipulation program to use for this sort of analysis. Free, widely supported and very capable.

http://rsbweb.nih.gov/ij/
User avatar
fooman
 
Posts: 50
Joined: Sun Dec 25, 2011 3:18 pm
Location: Wellington
Has Liked: 3 times
Has Been Liked: 29 times
Print view this post

#5)  Re: Photo Measurements

Postby Will Blozan » Thu Jan 10, 2013 10:29 pm

In my eastern time zone readjustmant stupor I did not see how you account for the diameter of the tree being 3+ feet behind the armspan.
User avatar
Will Blozan
 
Posts: 870
Joined: Fri Mar 12, 2010 8:13 pm
Location: North Carolina
Has Liked: 844 times
Has Been Liked: 310 times
Print view this post

#6)  Re: Photo Measurements

Postby dbhguru » Fri Jan 11, 2013 9:59 am

Will,

  I didn't. That was why I listed the sources of error and made the statement "The fact that the photographic result comes within 0.5 feet of the taped result has to be pure luck." My primary reason for these posts is to bring the topic up onto our collective radar screen. If you have an item of know size in an image and you have the distance to that object and each object, the diameter of which you wish to measure using Excel, then you can get well within the ball park for circular objects doing simple photographic analysis.

  I have just downloaded ImageJ per Matt's suggestion. There's a lot to learn, but as he says, the product is free and I can see that it has a big support group. It is well worth exploring. There are going to be countless opportunities to use our cameras as an extension of our lasers and clinometers. The approach that I take is to try to find relatively simple extensions of what we do, be they with mathematics or with equipment as opposed to taking on the challenge of very complex software systems that include literally thousands of features when all we're looking for is a few simple measurements and a protocol that involves only a few steps.  

  Now, more importantly, we want to hear about Alaska. Since we didn't further reports, I take it that you didn't make any big tree discoveries - and if you didn't, nobody could have. Next year, if you can plan on a Hawaii adventure with me, I guarantee you, the opportunities are limitless.

Bob
Robert T. Leverett
Co-founder and Executive Director
Native Native Tree Society
Co-founder and President
Friends of Mohawk Trail State Forest
User avatar
dbhguru
 
Posts: 3048
Joined: Mon Mar 08, 2010 10:34 pm
Location: Florence, Massachusetts
Has Liked: 0 times
Has Been Liked: 733 times
Print view this post

#7)  Re: Photo Measurements

Postby edfrank » Tue Jan 15, 2013 12:56 am

Bob,

I realize that people in all eras consisted of a wide variety of body sizes from tall to short, from long arms to short arms.  Have you found some table that might list average sizes of people at different times here in the US?  

I have searched on the internet and it is a morass of different changes over time and variations between different ethnic groups within the US.  I remember one old document I came across involving lumbering in PA or NY listed the workers as Whites, Indians, and Polocks.  Based upon different ethnic groups and living standards it might be expected that these different groups in this one location might have different proportions.

This pretty much summarized what I have found:

According to United States government data, the average height for U.S. adult males is 5 feet 9.9 inches, with the average female measuring 5 feet 4.3 inches. The average height of humans varies greatly across the world and depends on a number of factors including genetics and nutrition. It is difficult to come up with an exact worldwide average height because of variability in data collection between nations. Average human height has also fluctuated significantly throughout history.

Read more: The Average Height of Humans | eHow.com http://www.ehow.com/about_5462915_avera ... z2I16aFdZ9


Average height has oscillated throughout human history. The average height of a large population at a given moment in history often says a lot about the living conditions at that time. An Ohio State University study determined that European men during the 17th century averaged 5 feet 5 inches tall, whereas today they are a few inches taller. He attributes this to the abnormally cold weather and widespread political crisis in Europe during that century. The study also found that the Cheyenne people of North America were some of the tallest human beings ever to inhabit earth. He attributes this to the high protein, buffalo based diet of the Cheyenne.

Read more: The Average Height of Humans | eHow.com http://www.ehow.com/about_5462915_avera ... z2I17AYoMa


Here are some other links that have some useful information as opposed to the many that add little:

http://www.roymech.co.uk/Useful_Tables/Human/Human_sizes.html

http://www.macleans.ca/science/technology/article.jsp?content=20050404_103140_103140

http://www.quora.com/Human-Biology/Has-the-average-size-of-human-beings-changed-over-time

http://eh.net/encyclopedia/article/cuff.anthropometrics

The first link provides a table of anthropometric data.  Unfortunately there are some glitches in rows 32 and 33, I would assume that in row 32, span (arm span) they simply left the 1 off from in front of 810 and the number should be 1810.  Some of the numbers in row 33 are just wrong and I have no guesses as to what they might be.
.
"I love science and it pains me to think that so many are terrified of the subject or feel that choosing science means you cannot also choose compassion, or the arts, or be awe by nature. Science is not meant to cure us of mystery, but to reinvent and revigorate it." by Robert M. Sapolsky
User avatar
edfrank
Site Admin
 
Posts: 4100
{ IMAGES }: 0
Joined: Sun Mar 07, 2010 6:46 pm
Location: Reynoldsville, Pennsylvania, USA
Has Liked: 822 times
Has Been Liked: 600 times
Blog: View Blog (3)
Print view this post

#8)  Re: Photo Measurements

Postby Don » Wed Jan 16, 2013 4:16 pm

Ed/Bob-
Good comments!  It's always a good idea to have a way of relating to the subject in such photos, i.e., an object of known dimension, and in a stochastic way you've worked out a solution for unknown object dimensions.  

Another variability is within individual differences...most people's armspan is closely related to their height...since there are always exceptions, one occurs to me, in the guise of a champion boxer familiar to most of us of advanced wisdom and age (g).  Muhammed Ali (aka Cassius Clay) was deemed a more serious opponent because of his super-normal armspan (his height was 75", his reach 80"). I suppose that equal extremes in the other direction are likely.

Another comment that hasn't been mentioned (maybe for good reason?) has been the variability of scale in a photo due to the spherical nature of a lens and the flat nature of a photo. While in the center portion of a photo, this is a small distortion, but when the subject's proportion of the photo's extent is large, the distortion may be significant.

Otherwise, there is obvious merit to such photo measurement efforts.
-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
Don
 
Posts: 829
Joined: Tue Mar 09, 2010 1:42 am
Location: Anchorage, Alaska
Has Liked: 8 times
Has Been Liked: 104 times
Print view this post

#9)  Re: Photo Measurements

Postby edfrank » Wed Jan 16, 2013 4:35 pm

Don,

There are many layers of potential distortions in making measurements from old photos.  So to be used we need to try to figure out what ranges of human dimensions we are dealing with in each photo - and in most cases simply use an average with big error bars.   Distortion from the mechanics of the photograph and from things not being in the same plane also will add potential error to these estimates.  I wonder if these errors will add the the error bars significantly or simply be subsumed by the human size question to the point they are minimal?  They would be different on different photos depending on the distance to the camera, focal length, etc. So nothing was avoided "for good reason" it simply had not been considered yet.  I don't think we will ever be able to get the values very close, but should be able to get useful approximations on historical photos, and better results where we can actually measure the subject and have photo data.

Edward Frank
"I love science and it pains me to think that so many are terrified of the subject or feel that choosing science means you cannot also choose compassion, or the arts, or be awe by nature. Science is not meant to cure us of mystery, but to reinvent and revigorate it." by Robert M. Sapolsky
User avatar
edfrank
Site Admin
 
Posts: 4100
{ IMAGES }: 0
Joined: Sun Mar 07, 2010 6:46 pm
Location: Reynoldsville, Pennsylvania, USA
Has Liked: 822 times
Has Been Liked: 600 times
Blog: View Blog (3)
Print view this post

#10)  Re: Photo Measurements

Postby dbhguru » Wed Jan 16, 2013 6:27 pm

NTS,

   I continue chipping away at simple photo measurements. The following photo was from my iPhone camera. As you can see, shooting straight on, the target is measured photographically as being 11.4 inches in width. Its actual width is 11.25 inches.

               
                       
Screen shot 2013-01-16 at 4.53.45 PM.png
                                       
               


Bob
Robert T. Leverett
Co-founder and Executive Director
Native Native Tree Society
Co-founder and President
Friends of Mohawk Trail State Forest
User avatar
dbhguru
 
Posts: 3048
Joined: Mon Mar 08, 2010 10:34 pm
Location: Florence, Massachusetts
Has Liked: 0 times
Has Been Liked: 733 times
Print view this post

Next

Return to Measurement and Dendromorphometry

Who is online

Users browsing this forum: No registered users and 1 guest

cron