Wavy leaf surface area measurement

[Jess Hughes]

G'day,

Refer to the link below for images (colour copies or the originals with plain white background), as my attachments were too big.

https://picasaweb.google.com/115294069927812679719/Images03?authuser=0&feat=directlink

I was mainly wanting to work out the surface area for the attached leaves (new Muttonwood leaves.1.jpg), which it seems quite difficult to use the Image J software to do so normally, therefore I wonder whether it is indeed possible to account for wavy leaves (see attachment) by following the squished ridges with some kind of extra (automated) computed calculation. I imagine this would be an option that the person could select to analyse the leaf based on the type of leaf, i.e. variegated, wavy, spiky/round leaves (Australian sheoaks/bulloaks).

When it comes to using this software, it would be great if the person could select the type of information (final data results) they were interested in viewing i.e.

photosynthesis capability
Transpiration index
Surface area (showing the measurement unit, for instance mm^2, cm^2, inches^2, feet^2, m^2) Respiration capability (this may need to be based on other data above, perhaps an estimate with +/- figure).

I added all for images of the leaves I need to get surface area data for, which I have already individually cut out from A4 paper (80gsm) and measured on a digital scale and my calculated measurements are below.

Flat leaves - Blue and green background: Leaf area only, but results are for both A4 sheets combined (I would like data readings for each individual sheet).  
               
Weight of leaves (grams)                   6.5149
Leaf sample surface area (mm^2)     81436.25
Leaf sample surface area (m^2)        0.08143625

Wavy leaves - Fresh growth (Black background) and dirty leaves (red background)

                                                         Dirty leaves Fresh growth     Total
Weight of leaves (grams)                 1.231        2.518                 3.749
Leaf sample surface area (mm^2)   15387.5        31475                46862.5
Leaf sample surface area (m^2)      0.0153875 0.031475           0.0468625

It would be nice to know whether my data matches that which image J calculates for the surface area of the leaves, I also have the weight of the paper offcuts, but only for the the black (3.900g) and red (2.747g) background images and this should help to confirm the image processing and analyse capabilities/accuracy of Image J which I would then be able to include in my practical report on plant transpiration for univeristy.

Jesse

[Anderson, Charles (DNR)]

Getting total area of leaves in pixels in each image is easy enough:

run("8-bit");
setAutoThreshold("Default");
run("Convert to Mask");
run("Analyze Particles...", "size=0-Infinity circularity=0.00-1.00 show=Masks display exclude clear include summarize");

I assume you have some calibration images so you can convert pixels to mm^2.

No, you cannot measure the surface area of a wavy 3D leaf directly from a single 2D image. If you have some gold-standard by which you determine this for individual leaves, you could record image area and regress "known area" against image-measured area, then build some estimation model. You would have to take photographs without leaves touching or overlapping to reliably get individual leaf areas. You might try the watershed plugin to separate the leaves in some of the images you posted, but you will have to work through issues of broken or cracked leaves, what to do with petioles, etc.
 
Now, can you tell me how you photographed such uniform background colors (or have they already been enhanced in image analysis?).

Charles

-----Original Message-----
From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of Jess Hughes
Sent: Thursday, October 20, 2011 9:07 PM
To: [hidden email]
Subject: Wavy leaf surface area measurement

G'day,

Refer to the link below for images (colour copies or the originals with plain white background), as my attachments were too big.

https://picasaweb.google.com/115294069927812679719/Images03?authuser=0&feat=directlink

I was mainly wanting to work out the surface area for the attached leaves (new Muttonwood leaves.1.jpg), which it seems quite difficult to use the Image J software to do so normally, therefore I wonder whether it is indeed possible to account for wavy leaves (see attachment) by following the squished ridges with some kind of extra (automated) computed calculation. I imagine this would be an option that the person could select to analyse the leaf based on the type of leaf, i.e. variegated, wavy, spiky/round leaves (Australian sheoaks/bulloaks).

When it comes to using this software, it would be great if the person could select the type of information (final data results) they were interested in viewing i.e.

photosynthesis capability
Transpiration index
Surface area (showing the measurement unit, for instance mm^2, cm^2, inches^2, feet^2, m^2) Respiration capability (this may need to be based on other data above, perhaps an estimate with +/- figure).

I added all for images of the leaves I need to get surface area data for, which I have already individually cut out from A4 paper (80gsm) and measured on a digital scale and my calculated measurements are below.

Flat leaves - Blue and green background: Leaf area only, but results are for both A4 sheets combined (I would like data readings for each individual sheet).  
               
Weight of leaves (grams)                   6.5149
Leaf sample surface area (mm^2)     81436.25
Leaf sample surface area (m^2)        0.08143625

Wavy leaves - Fresh growth (Black background) and dirty leaves (red background)

                                                         Dirty leaves Fresh growth     Total
Weight of leaves (grams)                 1.231        2.518                 3.749
Leaf sample surface area (mm^2)   15387.5        31475                46862.5
Leaf sample surface area (m^2)      0.0153875 0.031475           0.0468625

It would be nice to know whether my data matches that which image J calculates for the surface area of the leaves, I also have the weight of the paper offcuts, but only for the the black (3.900g) and red (2.747g) background images and this should help to confirm the image processing and analyse capabilities/accuracy of Image J which I would then be able to include in my practical report on plant transpiration for univeristy.

Jesse

[Robert Dougherty]

Jesse and Charles,

On Oct 20, 2011, at 7:07 PM, Jess Hughes wrote:

>
> I was mainly wanting to work out the surface area for the attached leaves (new Muttonwood leaves.1.jpg), which it seems quite difficult to use the Image J software to do so normally, therefore I wonder whether it is indeed possible to account for wavy leaves (see attachment) by following the squished ridges with some kind of extra (automated) computed calculation. I imagine this would be an option that the person could select to analyse the leaf based on the type of leaf, i.e. variegated, wavy, spiky/round leaves (Australian sheoaks/bulloaks).


On Oct 21, 2011, at 6:41 AM, Anderson, Charles (DNR) wrote:

>> Getting total area of leaves in pixels in each image is easy enough:
>>
>> run("8-bit");
>> setAutoThreshold("Default");
>> run("Convert to Mask");
>> run("Analyze Particles...", "size=0-Infinity circularity=0.00-1.00 show=Masks display exclude clear include summarize");
>>
>> I assume you have some calibration images so you can convert pixels to mm^2.

Here is version I was working on when Charles' reply arrived:

1. Convert the images to binary.  With some of them, you can just use Process/Binary/Make Binary.  With others, you will need to set the threshold manually, perhaps using Image/Adjust/Color Threshold and then Process/Binary/Make Binary. Settings need to be found that cleanly separate the leaves from the spaces in between.
2. Notice whether the leaves are 0 or 255 in the binary image.  (It varies between the images.).  Move the mouse over a leaf and look at the pixel value in the status bar in the ImageJ toolbar.
3. Using Analyze/Set Measurements, select Area fraction.
4. Do Analyze/Measure.
5. Look at the area fraction in the Results Table.  Adjust this to reflect the area fraction of the leaves, as opposed to the space in between, by subtracting the value from 100 if necessary.  Refer to the results from step 2, and check to make sure the number looks right.
6. Divide by 100 and multiply by the area of the paper of whatever is covered by the image.

>>
>> No, you cannot measure the surface area of a wavy 3D leaf directly from a single 2D image. If you have some gold-standard by which you determine this for individual leaves, you could record image area and regress "known area" against image-measured area, then build some estimation model. You would have to take photographs without leaves touching or overlapping to reliably get individual leaf areas. You might try the watershed plugin to separate the leaves in some of the images you posted, but you will have to work through issues of broken or cracked leaves, what to do with petioles, etc.

As Charles pointed out, this process will give a 2D projected are, not a 3D surface area reflecting non-planar leaves. If your are not holding them flat already by using a scanner with a cover, you might consider a glass cover plate or at least applying sufficient  tension with the tape that you seem to be using.  A perforated plate with suction would be another flattening technology. 3D measurement techniques exist, but don't seem necessary here.

>
> When it comes to using this software, it would be great if the person could select the type of information (final data results) they were interested in viewing i.e.
>
> photosynthesis capability
> Transpiration index
> Surface area (showing the measurement unit, for instance mm^2, cm^2, inches^2, feet^2, m^2) Respiration capability (this may need to be based on other data above, perhaps an estimate with +/- figure).

This would require a specialized plugin or macro to be created for ImageJ.

Bob


Robert Dougherty, Ph.D.
President, OptiNav, Inc.
1414 127th Place NE #106
Bellevue, WA 98005
Tel. [phone]
FAX [phone]
www.optinav.com
[hidden email]