I have images of erect grasses and sedge in a wind tunnel (2 example pictures attached) and I need to get the surface area of those plants. All my pictures are taken upwind in a wind tunnel and prior to any manipulations such that the the plant matter represents the surface area available to catch wind blown sand that the plants later get blasted by. However, I was unable to take the pictures with a white/no background and so there is a lot of background noise in terms of the other objects that were in the wind tunnel. Does anyone know how to remove all but one color in imageJ? I thought that this would be a very straight forward thing to do - select a green color range and remove all else, but I am struggling to figure out how to do this. In the past, with white backgrounded images and leaves, I have converted to binary such that all the plant matter was black and all else white such that I can get the SA of the black. However, I am having trouble getting to a point to where I could convert to binary and only have my plant matter be black. I'd appreciate any help! Thanks for your time!
Dear Bianca
Are you working with any Image Processing program ? (like MAtlab, LabVIEW) if not its a must.
Just to demonstrate I sent you the images filtered to 3 and 4 color levels (I did in matlab) but you can count the number of green pixels from the image matrix after digitizing.
Regards
Moshe
Hi Moshe,
Thank you so much for responding to my question. No I am currently not using any imaging processing program like MAtlab or labVIEW. I had hoped to use ImageJ for these analyses. I think that you can do what you are suggesting and have done here. In image J if you go image--> type--> 8-bit color--> select number of colors from 2-256. If I do this and select 4 colors for 7 then my image looks close to yours. However, I am unsure of next steps from here for how you count the green pixels. I'd appreciate any further guidance as to where to go from here to further digitize.
Hello Bianca,
I am quite sure your software is doing the same as Matlab, The logic of plotting the image should be the same.
The Matlab function F-imread(file) is digitizing the image to a NxM size matrix . Look for similar function inn your software.
Each cell of the matrix contains the digitized color value. So you will have to identify the value for green and count the total number of pixels.
It may sounds easier to say than do but I am sure with a bit of trial and error attempts you will get the skill. In order to prove your method you will have to zero all the rest of the pixels and plot the modified image to get the original shape of the plants.
You could ease your analysis if you hang a real white sheet in the background each time you take a snapshot. This may also prove your method.
Hoping it'll help
Moshe
I don't know as the answer was posted a year ago, I guess by now you probably found a solution. You can split channels in green, red and blue and see what you can get out of it.
Image -> color -> split channels
- Badges
- Science topic
- Similar topics
- Chemistry
- Organic Chemistry