Select an area with intensity color

1 view (last 30 days)
Frank
Frank on 24 Mar 2014
Commented: Image Analyst on 25 Mar 2014
Hi, I'm trying to select the area selected in red from the picture below. For now, I'm getting the area in green. The original picture is in RGB color and I transform it to intensity image. Then, I use the functions adapthisteq and imcontrast to improve the contrast. Then, I transform my image to binary image and I find the "true" pixels.
My problem is that the area selected in red is too light compared to the whole image. I tried to work with my contrast options but I can't get that area except if I improve my level criteria for my binary conversion. I want to make it automatic so the user dont have to do something.

Answers (1)

Image Analyst
Image Analyst on 24 Mar 2014
How about using imfreehand()? Does the area always have smooth borders like the red region? Or does it go out to the straight edges that seem to sort of border it? How did you decide on the thresholds that took the dark blobs? Why not adjust it to take the lighter areas?
Try my interactive thresholding app to see if there is a global threshold that will even work at all. http://www.mathworks.com/matlabcentral/fileexchange/?term=authorid%3A31862
  2 Comments
Image Analyst
Image Analyst on 25 Mar 2014
Frank incorrectly posted his comment to me as an "Answer" so I moved it to here:
I could use imfreehand() but my program will work with many images (>50), so it would be pretty long. I'm looking for a more automatic way of doing it.
For now, I chose the level by looking at my results and I try to find the best one. If I change it to a higher level, it will select my light fibers.
I tried your interactive thresholding app but I don't think I can find a single level that will fit to all the images.
Image Analyst
Image Analyst on 25 Mar 2014
If there is no single global threshold that will work, then you will not be able to find one by looking at the results. Basically that's what my thresholding app does - it lets you look at every single threshold value to find the best one. You will have to do other processing before thresholding since the image is not yet ready for the thresholding step.

Sign in to comment.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!