Hi Question is about eleminating unwanted regions in binary image.

I work on fundus image , bwareaopen() function removes all connected components fewer than P pixels. But in my image I need to remove all the regions which have more than 11 pixels how can I do so? And how can I calculate the metrics of each region and eliminate the structures having metrics less than 0.95?

Answer 1

[Image Analyst]

So P=11, right? Did you try this:

binaryImage = bwareaopen(binaryImage, 11);

Regarding "metrics", I think you need to do a spatial calibration, and I attach my demo for that. So then use regionprops() and multiply linear measurements by the spatial calibration factor and area measurements by the spatial calibration factor squared.

[Pradeep Gowda]

If I use p=11 it will eliminate all the regions that have less than 11 pixels but I need to eliminate all the regions that have more than 11 pixels .

[Image Analyst]

Try bwareafilt:

smallBlobs = bwareafilt(binaryImage,[1, 11]);

[Pradeep Gowda]

If I'm not wrong that function is only available in r2015b . I work with r2013b . so any other suggestions?

[Meghana Dinesh]

You can do what Image Analyst mentioned:

 binaryImage_out = bwareaopen(binaryImage_in, 11);

Later, subtract binaryImage_in and binaryImage_out.

[Image Analyst]

binaryImage = binaryImage - bwareaopen(binaryImage, 11);