connected components (or just regionprops) on integer image

Hi all --

I am attempting to segment an image and then gather statistics (regionprops) on each image object. I have completed the segmentation and there are 110 unique values that are mostly (but not entirely) grouped spatially. My question is: How do I conduct a connected components analysis on this non-binary output? I'd like to avoid running watershed on this because it seems unnecessary to introduce the 0-value borders (thus reducing image information content). Any suggestions?

Thanks, Mike

Answer 1

[Matt J]

I would do something like this,

vals=unique(yourImage(:));

fun = @(v) regionprops(yourImage==v,...); 

stats = cell2mat( arrayfun( fun  , vals, 'uni',0)   );