Comments on: A new look for connected component labeling in R2009a – Part 2 http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/ Steve Eddins manages the Image & Geospatial development team at The MathWorks and coauthored Digital Image Processing Using MATLAB. He writes here about image processing concepts, algorithm implementations, and MATLAB. Sat, 11 Feb 2012 18:27:50 +0000 hourly 1 http://wordpress.org/?v=3.2.1 By: Nick http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-24784 Nick Mon, 09 Jan 2012 15:54:46 +0000 http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-24784 Thanks! Thanks!

]]> By: Steve Eddins http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-24780 Steve Eddins Wed, 04 Jan 2012 21:07:34 +0000 http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-24780 Nick—bwselect might be useful. Nick—bwselect might be useful.

]]> By: Nick http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-24779 Nick Wed, 04 Jan 2012 20:15:02 +0000 http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-24779 Hi Steve, I am attempting to find the area of a small structure in a thresholded image, and I was hoping to use the connected component labeling properties of |bwconncomp| to do so. Is there a way to feed in coordinates (likely specified with |ginput|) and ask |bwconncomp| to only focus on that particular region? I would essentially like to do the reverse of the example provided in the |bwlabel| documentation: [r, c] = find(bwlabel(BW)==2) that is, to specify the coordinates and only label that region. I see that functions like |regionprops| and |bwconncomp| work on the entire image, and I wonder if there is a more sophisticated way to only focus on a certain region than to create a duplicate image cropped to the region of interest in the background. Thanks for your help! Hi Steve,

I am attempting to find the area of a small structure in a thresholded image, and I was hoping to use the connected component labeling properties of |bwconncomp| to do so. Is there a way to feed in coordinates (likely specified with |ginput|) and ask |bwconncomp| to only focus on that particular region?

I would essentially like to do the reverse of the example provided in the |bwlabel| documentation:

[r, c] = find(bwlabel(BW)==2)

that is, to specify the coordinates and only label that region.

I see that functions like |regionprops| and |bwconncomp| work on the entire image, and I wonder if there is a more sophisticated way to only focus on a certain region than to create a duplicate image cropped to the region of interest in the background.

Thanks for your help!

]]> By: Steve http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-24523 Steve Mon, 10 Oct 2011 13:22:54 +0000 http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-24523 Shilan—Try this: <pre class="code"> s = regionprops(bw, 'area', 'centroid'); a = [s.Area]; s = s(a > 100); </pre> Shilan—Try this:

s = regionprops(bw, 'area', 'centroid');
a = [s.Area];
s = s(a > 100);
]]> By: shilan http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-24503 shilan Mon, 26 Sep 2011 11:59:25 +0000 http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-24503 using "bwconncomp" function, I want to determine centeroid of regions with a size greater than for example 100! how is it possible to combine this function with something like find? Just in a way, for instance we use <pre> [r, c] = find(bwlabel(BW)==2) </pre> or there is another way? using “bwconncomp” function, I want to determine centeroid of regions with a size greater than for example 100! how is it possible to combine this function with something like find? Just in a way, for instance we use

[r, c] = find(bwlabel(BW)==2)

or there is another way?

]]> By: Steve http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-23826 Steve Wed, 05 Jan 2011 13:47:23 +0000 http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-23826 Ramesh—regionprops on a 256-by-256 test image containing 88 connected components runs in 10 ms on my computer. Do you regard this as very very slow? Ramesh—regionprops on a 256-by-256 test image containing 88 connected components runs in 10 ms on my computer. Do you regard this as very very slow?

]]> By: Ramesh http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-23824 Ramesh Mon, 03 Jan 2011 06:19:11 +0000 http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-23824 Steve, Regionprops runs very very slowly when I use it as a part of a code wherein I need to do labelling (of regions) on a set of images. I use Matlab v7.8.0 R2009a Any tips on how to speed it up? Thanks Steve,

Regionprops runs very very slowly when I use it as a part of a code wherein I need to do labelling (of regions) on a set of images. I use Matlab v7.8.0 R2009a

Any tips on how to speed it up?

Thanks

]]> By: Steve http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-23615 Steve Thu, 14 Oct 2010 20:16:14 +0000 http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-23615 Tanya—I recommend reinstalling. If that doesn't work then I recommend contacting tech support. Tanya—I recommend reinstalling. If that doesn’t work then I recommend contacting tech support.

]]> By: Tanya http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-23614 Tanya Thu, 14 Oct 2010 20:13:37 +0000 http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-23614 Steve- I am currently running R2010a with the Image Processing Toolbox that came with it installed (Academic license). Should I try re-installing everything, or contact tech support to help, or ? Steve- I am currently running R2010a with the Image Processing Toolbox that came with it installed (Academic license). Should I try re-installing everything, or contact tech support to help, or ?

]]> By: Steve http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-23613 Steve Thu, 14 Oct 2010 19:53:29 +0000 http://blogs.mathworks.com/steve/2009/07/13/a-new-look-for-connected-component-labeling-in-r2009a-part-2/#comment-23613 Tanya—You do not have a valid installation of Image Processing Toolbox for the platform you are running MATLAB on. Tanya—You do not have a valid installation of Image Processing Toolbox for the platform you are running MATLAB on.

]]>