http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/ 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 http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-23001 Steve Fri, 16 Apr 2010 17:53:31 +0000 http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-23001 Aaron—The function <tt>bwlabeln</tt> doesn't use <tt>dmperm</tt>. It uses an implementation of the union-find algorithm. The implementation was based on the union-find description in Robert Sedgewick, Algorithms: Parts 1–4, 3rd edition, Addison-Wesley, 1998, pp. 10–23. Aaron—The function bwlabeln doesn’t use dmperm. It uses an implementation of the union-find algorithm. The implementation was based on the union-find description in Robert Sedgewick, Algorithms: Parts 1–4, 3rd edition, Addison-Wesley, 1998, pp. 10–23.

]]> http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-23000 Aaron Thode Fri, 16 Apr 2010 13:57:15 +0000 http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-23000 Steve, thank you for your reply. Last question: I'm preparing a scientific manuscript that used bwlabeln extensively. The journal discourages web citations, particularly blogs (I know, I know...), unless more formal references are unavailable, so out of the references you've sprinkled throughout this series, are there one or two you'd recommend as appropriate for further information on the particular component connecting technique that uses Dulmage-Mendelsohn permutations? You've already posted a reference for run-length encoding here...thanks! Steve, thank you for your reply. Last question: I’m preparing a scientific manuscript that used bwlabeln extensively. The journal discourages web citations, particularly blogs (I know, I know…), unless more formal references are unavailable, so out of the references you’ve sprinkled throughout this series, are there one or two you’d recommend as appropriate for further information on the particular component connecting technique that uses Dulmage-Mendelsohn permutations? You’ve already posted a reference for run-length encoding here…thanks!

]]> http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-22997 Steve Thu, 15 Apr 2010 12:15:10 +0000 http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-22997 Aaron—The advantage of bwconncomp is that it typically uses a lot less memory for its output. bwlabel produces a label matrix in double-precision format with the same size as the input image. That can be a lot of wasted memory, particularly when there aren't many foreground pixels relative to the total number of image pixels. Aaron—The advantage of bwconncomp is that it typically uses a lot less memory for its output. bwlabel produces a label matrix in double-precision format with the same size as the input image. That can be a lot of wasted memory, particularly when there aren’t many foreground pixels relative to the total number of image pixels.

]]> http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-22995 Aaron Thode Wed, 14 Apr 2010 21:09:30 +0000 http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-22995 Greetings Steve, I just stumbled across your blog when researching bwlabel, and found it extremely clear and useful. Would you mind commenting how bwconncomp differs from bwlabel? Seems like run-length encoding and dmperm are still used, so why is it more efficient? Greetings Steve, I just stumbled across your blog when researching bwlabel, and found it extremely clear and useful. Would you mind commenting how bwconncomp differs from bwlabel? Seems like run-length encoding and dmperm are still used, so why is it more efficient?

]]> http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-22685 João Henriques Tue, 09 Feb 2010 18:12:30 +0000 http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-22685 I'd just like to say that I also finished reading this series, and found it very informative! Thank you for sharing! I’d just like to say that I also finished reading this series, and found it very informative! Thank you for sharing!

]]> http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-21711 Steve Mon, 27 Apr 2009 00:14:51 +0000 http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-21711 Ashwati—Thanks! I forwarded your comment to Dr. Gonzalez, who wrote the frequency domain processing chapter. Ashwati—Thanks! I forwarded your comment to Dr. Gonzalez, who wrote the frequency domain processing chapter.

]]> http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-21710 Ashwati Sun, 26 Apr 2009 13:58:09 +0000 http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-21710 Dear Steve, I am currently doing a project in Digital Image Processing. I just wanted to say that I have followed your book, DIP with Matlab, and have found it a very satisfying experience, especially the section of Frequency Domain processing, cleared a lot of cobwebs. Thank you very much. Ashwati Dear Steve,
I am currently doing a project in Digital Image Processing. I just wanted to say that I have followed your book, DIP with Matlab, and have found it a very satisfying experience, especially the section of Frequency Domain processing, cleared a lot of cobwebs. Thank you very much.
Ashwati

]]> http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-21196 Steve Wed, 29 Oct 2008 22:47:02 +0000 http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-21196 Jxwht—Not directly by using indexing offsets, but it can be programmed in MATLAB. I have done it in the past, but a long time ago. I don't have code for it anymore. Jxwht—Not directly by using indexing offsets, but it can be programmed in MATLAB. I have done it in the past, but a long time ago. I don’t have code for it anymore.

]]> http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-21179 jxwht Tue, 21 Oct 2008 12:23:32 +0000 http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-21179 Hi Steve, As you mentioned before, we can find the 4- or 8- connected "pairs" by using linear indexing and linear neighbor index offsets, is there a similar algorithm to find the 4- or 8- connected "pairs" RUNs? Thank you! Hi Steve,
As you mentioned before, we can find the 4- or 8- connected “pairs” by using linear indexing and linear neighbor index offsets, is there a similar algorithm to find the 4- or 8- connected “pairs” RUNs?
Thank you!

]]> http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-14131 Steve Mon, 05 Nov 2007 23:52:28 +0000 http://blogs.mathworks.com/steve/2007/06/12/connected-component-labeling-part-7/#comment-14131 Emanuele—I'm glad you liked it. Thanks for taking the time to comment. Emanuele—I’m glad you liked it. Thanks for taking the time to comment.

]]>