Pixel extraction and Reclassify

I had segmented image and want to detect of these pixels by extracting all pixels that do not have the same label in their 3×3 neighborhood (N_ns), and then re-classify of these extracted pixels using local information in a 5×5 neighborhood (N_rl). What can I do? Thanks..

Answer 1

[Image Analyst]

labelsYouWantToKeep = [3,5,9]; % Whatever blob numbers you want.
% Get a new labeled image with only those blobs in it.
newLabeledImage = ismember(originalLabeledImage, labelsYouWantToKeep);
imshow(newLabeledImage, []);  % Display it.
% Remeasure with only those blobs in the image.
newMeasurements = regionprops(newLabeledImage);

[dian_pratama]

It's work. Thanks IA, U fix my problems. But, I'm still not understand: [3, 5 and 9] refers to what?

[Image Analyst]

Well let's say you have 100 blobs in your binary image. What numbered blobs do you want to keep? If you don't want blobs #3, 5, and 9, like in my example, then which ones DO you want? There's no way that I could know which blobs you are thinking of.

[dian_pratama]

How about N_ns and N_rl? Can I apply that in matlab? I think before, #3 is 3x3 neighborhood, #5 is 5x5 neighborhood, and #9 is image size. LOL..

[Image Analyst]

Explain more of what you want to do, especially about what you want to do with blobs based on what is in their "neighborhood". Say blob #15 and blob #27 have their centroids or perimeters within 9 pixels of each other - now what????