Objective Assessment of Area of Algae Eaten by Mayflies

[Pete Cadmus]

I am an ecologist trying to assess the amount of periphytic (surrounding rocks) algae that is eaten by mayflies.  I have cultured algae on 7/8” by 7/8” tiles.  I hope to photograph the tiles before the mayflies graze on them and after.  Then use ImageJ to asses the surface area of algae compared to the surface area that was grazed free of algae.  Unfortunately the algae is not able to be grown uniformly.  Algae can appear as thick dark green filaments to only slightly green/brown tinted film.  The tile appears as white.  Often where the insect grazes algae only a slight color difference exists between the white tile and very thin film of diatoms.  Also, I am working in a greenhouse so photos will have inconsistent lighting from day to day even under a hood.  

I have been reading your archive all day, I have learned a great deal, you are all doing some impressive work.  Can someone point me in the right direction as to what plug-ins to install and what process to follow.  What approach do you feel is the best, most objective way to asses the area grazed?

Thanks for your time,
Pete Cadmus
Graduate Student of Fish, Wildlife and Conservation Biology
Colorado State University
petecadmus( a  t ) hot mail (d o t) com
970-491-5563 lab
970-482-0784 home

P.S. I am happy to send sample photos of algae by email

[Pete Cadmus]

Any one?  Hello...

The best i have come up with is the following:
change RGB photo to 16bit BW photos.  Use rotate and crop to show only the face of the tile.  Set a scale to the image using the known width of the tile.  Use threshold adjuster to select only the tile or only the algae.  Then select measure and receive the area selected by the threshold adjuster.

Problems include:
1) sometimes the threshold sliders only let me select things lighter than a color and sometimes is switches to darker. I can't figure out why
2) i am just arbitrarily sliding the threshold adjuster till it "looks about right"  this is not objective.
3) THE BIG PROBLEM pixels that i know are tile (not green) are selected as algae just because it is a little shaded or darker than the tile. I have tried flash bulbs, underwater cameras, ring shaped florescent lights, black hoods to block the sun...  In every case it seams that one side is slightly darker on one side than the other.  Also i seams that pixils near thick algae are selected as algae even though they are obviously bare tile. ( Matlab did the same thing)

Any way to fix these issues?

Any way to eliminate the shaded bare tile pixels being mistaken for light green algae pixels by incorporating the green RBG value?

Any way to assign a value something like units of darkness above a threshold for each pixel.  Then sum those units for all pixels.  Take many observations, growing maybe a hundred tiles.  Then plot this crazy sumation of dark units against the grams of algae scraped off each corresponding tile and fit a curve to it.  Then use this equation to approximate the grams of algae on the tile.  Is this possible, ohh great ImageJ gurus?


Thanks for your time,
Pete Cadmus
Graduate Student of Fish, Wildlife and Conservation Biology
Colorado State University
petecadmus( a  t ) hot mail (d o t) com
970-491-5563 lab
970-482-0784 home