Vegetative cover estimates- grid overlay possible?

[sarah Al-beidh]

Hi all,

A few months ago I emailed the list to ask how I could calculate the proportion of vegetation on hundreds of 3mx3m plot photographs.  Your suggestions were to adjust the color threshold and then threshold each image and finally to use a macro to calculate the proportion of white pixels on each thresholded photo.  This worked a treat but it has ended up meaning that by standardising the protocol (ie. hue=32 etc.) I end up underestimating vegetative cover on many of the plots (where vegetation is either pure white or dark brown).

Consequently, we have come up with an alternative idea that would involve overlaying a grid over each photograph and counting the number of filled squares to gain a separate estimate of vegetative cover per plot.  Is there any way of overlaying a grid of a set size on each photo using Image J?

Many thanks,
Sarah

Dr. Sarah Al-Beidh
RHS Wisley

[Norbert Vischer-2]

I want to distribute my plugin as a .jar file. However, I noticed that the file downloaded by IE 8 can be smaller than the original (eg 1 MB instead 1.4MB), in which case it is corrupted.
In the dialog, the .jar file is classified as "unknown file type" and gives me the choice "download as executable Jar File", or "All Files". In either case, the system reports "download complete" - no error whatsoever.

This may happen on IE 8 + Windows 7 + Parallels, but occasionally also on a IE 8 + Windows 7 + Dell.
Using other jar files, or using Firefox, or using a Mac: everything is ok.


Norbert Vischer

[BenTupper]

Hi,

On Feb 23, 2011, at 4:26 AM, sarah Al-beidh wrote:

> Consequently, we have come up with an alternative idea that would  
> involve overlaying a grid over each photograph and counting the  
> number of filled squares to gain a separate estimate of vegetative  
> cover per plot.  Is there any way of overlaying a grid of a set size  
> on each photo using Image J?
>

Check out http://rsb.info.nih.gov/ij/plugins/grid.html

Cheers,
Ben

[sarah Al-beidh]

Hi Ben,
That works.  The only problem is that the grid is centred in the middle so that the edge of the photograph has half squares all around it.  I need the grid to start at the edges so that there are 10 full squares along the width and the same along the height.  Is there any way to adjust the plugin to mean that the edges of squares start at each edge?

Many thanks
Sarah

________________________________________
From: ImageJ Interest Group [[hidden email]] On Behalf Of Ben Tupper [[hidden email]]
Sent: 23 February 2011 15:39
To: [hidden email]
Subject: Re: Vegetative cover estimates- grid overlay possible?

Hi,

On Feb 23, 2011, at 4:26 AM, sarah Al-beidh wrote:

> Consequently, we have come up with an alternative idea that would
> involve overlaying a grid over each photograph and counting the
> number of filled squares to gain a separate estimate of vegetative
> cover per plot.  Is there any way of overlaying a grid of a set size
> on each photo using Image J?
>

Check out http://rsb.info.nih.gov/ij/plugins/grid.html

Cheers,
Ben

[Glen MacDonald-2]

Dear Sarah,
I'm sending you an ImageJ macro that I've been recycling for different stereology applications.  Note that the scaling on this version is only a 'placeholder'  for scanned EM negatives and has not yet been calibrated - hoping to get the calibration images this week.  But you can either comment out that part of the macro or reset the scaling in Image>Properties after running the setup command.  I'm not sending here because the scaling is not yet verified.  

Alternatively, you can edit the Grids plugin within ImageJ to change the the appearance of the grid.  

Regards,
Glen

Glen MacDonald
Core for Communication Research
Virginia Merrill Bloedel Hearing Research Center
Box 357923
University of Washington
Seattle, WA 98195-7923  USA
[phone]
[hidden email]








On Feb 23, 2011, at 1:26 AM, sarah Al-beidh wrote:

> Hi all,
>
> A few months ago I emailed the list to ask how I could calculate the proportion of vegetation on hundreds of 3mx3m plot photographs.  Your suggestions were to adjust the color threshold and then threshold each image and finally to use a macro to calculate the proportion of white pixels on each thresholded photo.  This worked a treat but it has ended up meaning that by standardising the protocol (ie. hue=32 etc.) I end up underestimating vegetative cover on many of the plots (where vegetation is either pure white or dark brown).
>
> Consequently, we have come up with an alternative idea that would involve overlaying a grid over each photograph and counting the number of filled squares to gain a separate estimate of vegetative cover per plot.  Is there any way of overlaying a grid of a set size on each photo using Image J?
>
> Many thanks,
> Sarah
>
> Dr. Sarah Al-Beidh
> RHS Wisley

[Rasband, Wayne (NIH/NIMH) [E]]

On Feb 23, 2011, at 11:03 AM, sarah Al-beidh wrote:

> Hi Ben,
> That works.  The only problem is that the grid is centred in the middle so that the edge of the photograph has half squares all around it.  I need the grid to start at the edges so that there are 10 full squares along the width and the same along the height.  Is there any way to adjust the plugin to mean that the edges of squares start at each edge?

You can use the Grid Overlay plugin at

   http://imagej.nih.gov/ij/plugins/graphic-overlay.html

-wayne

> ________________________________________
> From: ImageJ Interest Group [[hidden email]] On Behalf Of Ben Tupper [[hidden email]]
> Sent: 23 February 2011 15:39
> To: [hidden email]
> Subject: Re: Vegetative cover estimates- grid overlay possible?
>
> Hi,
>
> On Feb 23, 2011, at 4:26 AM, sarah Al-beidh wrote:
>
>> Consequently, we have come up with an alternative idea that would
>> involve overlaying a grid over each photograph and counting the
>> number of filled squares to gain a separate estimate of vegetative
>> cover per plot.  Is there any way of overlaying a grid of a set size
>> on each photo using Image J?
>>
>
> Check out http://rsb.info.nih.gov/ij/plugins/grid.html
>
> Cheers,
> Ben<grid +photo.jpg>

[sarah Al-beidh]

Just a quick thank- you to Ben, Glen and Wayne.
Your helpful advice has been very much appreciated!

Sarah

[dvor]

Hi Sarah

Using only filled grids to estimate the area covered will lead to an underestimate of the vegetation cover.

In principle, one can use e.g. one corner only (say, the upper left "point") or the centroid of each grid square (or rectangle!) to measure the coverage. The number of e.g. centroids hitting vegetation divided by the total number of centroids hitting the image gives the % coverage. The number of points x the calibrated area per grid square gives the area.

It is not necessary to have the grids aligned with the image -- they do need to contain the image areas of interest, though (so having half squares around an image is okay). In fact, the grid should be randomly positioned on the image to get best results.

This is the stereological procedure commonly called "point counting".

Dvoralai


>Hi Ben,
>That works.  The only problem is that the grid is centred in
 the middle so that the edge of >the photograph has half squares all
around it.  I need the grid to start at the edges so that >there are 10
full squares along the width and the same along the height.  Is there
any way >to adjust the plugin to mean that the edges of squares start at
each edge?
>
>Many thanks
>Sarah
>

[sarah Al-beidh]

Hi Dvoralai,

Many thanks.  Yep- I have used pin frames in the field based on the same principle- i.e. number of touches of the vegetation along a pin lowered into the vegetation gives an idea of the density of vegetation at that position.  

I originally tried to be even more accurate than this by converting my coloured images to black and white where the vegetation was one colour and everything else, the other.  I then calculated the number of pixels of vegetation (e.g. white) out of the total number of pixels in the transformed image (black +white) to create a measure of vegetative cover. However, that method underestimated vegetation because it was hard to include vegetation of different colours (brown, green, white, purple etc), without also including some unwanted (non-vegetative) parts of the picture as well.  There doesn't seem to be a perfect, easy method!

Using grid squares is sufficiently accurate for my purposes- in fact this measure is simply going to be used as a covariate in an analysis and more important than the actual values of the vegetative cover are the relative values across multiple plots and multiple seasons/ recording events.  


Many thanks for your suggestion!
Sarah

-----Original Message-----
From: ImageJ Interest Group [mailto:[hidden email]] On Behalf Of dvor w
Sent: 24 February 2011 14:34
To: [hidden email]
Subject: Re: Vegetative cover estimates- grid overlay possible?

Hi Sarah

Using only filled grids to estimate the area covered will lead to an underestimate of the vegetation cover.

In principle, one can use e.g. one corner only (say, the upper left "point") or the centroid of each grid square (or rectangle!) to measure the coverage. The number of e.g. centroids hitting vegetation divided by the total number of centroids hitting the image gives the % coverage. The number of points x the calibrated area per grid square gives the area.

It is not necessary to have the grids aligned with the image -- they do need to contain the image areas of interest, though (so having half squares around an image is okay). In fact, the grid should be randomly positioned on the image to get best results.


This is the stereological procedure commonly called "point counting".

Dvoralai


>Hi Ben,
>That works.  The only problem is that the grid is centred in
 the middle so that the edge of >the photograph has half squares all
around it.  I need the grid to start at the edges so that >there are 10
full squares along the width and the same along the height.  Is there
any way >to adjust the plugin to mean that the edges of squares start at
each edge?
>
>Many thanks
>Sarah
>