Object extraction from background
Show older comments
Hello, I have so many image that i need to extract object from them(remove the background) i have tested automatic thresholding to estimate the background gray level, also i hae tried Level set , but both method give good result for some images and poor result for other , i'm asking if there is a robust and good way to do this task. Please can anyone put me in the right direction.
here is an example that contain many object:

here is another one that contain one object:

Thank you
Answers (2)
Image Analyst
on 31 Jan 2017
Edited: Image Analyst
on 31 Jan 2017
0 votes
I'd probably do a texture filter like stdfilt(), maybe ORed with an edge detector, and then threshold and fill the blobs with imfill. Let me know if you can't figure it out. Attach the images without the red graphical outlines on them.
1 Comment
Image Analyst
on 1 Feb 2017
0 votes
See attached demo. It could be improved, but for 10 minutes work, it's a start.

10 Comments
Lafi Raed
on 1 Feb 2017
Image Analyst
on 1 Feb 2017
Have you observed that the best threshold changes for different images? If so you'll have to use some kind of rule to pick the threshold instead of using a fixed one. One I like is the triangle threshold, attached.
Image Analyst
on 2 Feb 2017
Yes - you can use splines or activecontour(). See attached demos.
Lafi Raed
on 3 Feb 2017
Image Analyst
on 3 Feb 2017
I've given links for that before: http://www.mathworks.com/matlabcentral/answers/?search_submit=answers&query=tag%3Ashadow&term=tag%3Ashadow
Image Analyst
on 4 Feb 2017
Sorry, I don't and can't recommend one. The one I saw at the University of Dayton, that they developed, looked pretty good.
Lafi Raed
on 4 Feb 2017
Image Analyst
on 4 Feb 2017
Their research page is here: https://www.udayton.edu/engineering/centers/vision_lab/research-pages/index.php
One of them is about shadow removal.
Categories
Find more on Image Segmentation in Help Center and File Exchange
Products
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!