Comments on: Nonflat grayscale dilation and erosion http://blogs.mathworks.com/steve/2006/10/23/nonflat-grayscale-dilation-and-erosion/ 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. Wed, 01 Feb 2012 13:58:45 +0000 hourly 1 http://wordpress.org/?v=3.2.1 By: Steve http://blogs.mathworks.com/steve/2006/10/23/nonflat-grayscale-dilation-and-erosion/#comment-24240 Steve Thu, 26 May 2011 17:12:01 +0000 http://blogs.mathworks.com/steve/?p=91#comment-24240 Dave—There's this option: <pre class="code"> se = strel('ball', R, H); </pre> What other shapes are you interested in? Dave—There’s this option:

se = strel('ball', R, H);

What other shapes are you interested in?

]]> By: Dave http://blogs.mathworks.com/steve/2006/10/23/nonflat-grayscale-dilation-and-erosion/#comment-24239 Dave Thu, 26 May 2011 17:00:35 +0000 http://blogs.mathworks.com/steve/?p=91#comment-24239 Sorry, I'm eroding with my strel(). That should make more sense! Sorry, I’m eroding with my strel(). That should make more sense!

]]> By: Dave http://blogs.mathworks.com/steve/2006/10/23/nonflat-grayscale-dilation-and-erosion/#comment-24238 Dave Thu, 26 May 2011 16:52:05 +0000 http://blogs.mathworks.com/steve/?p=91#comment-24238 I'm using a non-flat strel() to subsample a volume of interest in a 3D data set. I'm using the element to dilate a VOI mask to give me the valid top-left-front corner subscripts that are at least some distance X from all other subsamples and still within the VOI. The nice thing about dilating a binary mask is you don't have to mess with the height values. I can see an advantage to passing a function to strel() to compute the heights, instead of a user-computed matrix. Perhaps in the next release? :) I’m using a non-flat strel() to subsample a volume of interest in a 3D data set. I’m using the element to dilate a VOI mask to give me the valid top-left-front corner subscripts that are at least some distance X from all other subsamples and still within the VOI.

The nice thing about dilating a binary mask is you don’t have to mess with the height values. I can see an advantage to passing a function to strel() to compute the heights, instead of a user-computed matrix. Perhaps in the next release? :)

]]> By: Steve http://blogs.mathworks.com/steve/2006/10/23/nonflat-grayscale-dilation-and-erosion/#comment-23569 Steve Tue, 05 Oct 2010 15:32:01 +0000 http://blogs.mathworks.com/steve/?p=91#comment-23569 Rod—Gray-scale dilation is most commonly performed with a flat, zero-height structuring element. For that common case it is basically equivalent to binary image dilation in that both can be considered to be local-max operations. Rod—Gray-scale dilation is most commonly performed with a flat, zero-height structuring element. For that common case it is basically equivalent to binary image dilation in that both can be considered to be local-max operations.

]]> By: Rod http://blogs.mathworks.com/steve/2006/10/23/nonflat-grayscale-dilation-and-erosion/#comment-23568 Rod Tue, 05 Oct 2010 15:21:38 +0000 http://blogs.mathworks.com/steve/?p=91#comment-23568 Wouldn't dilating a binary image be quite different from dilating a gray scale image? The results from dilating the gray scale image would be dependent on the scale range of the gray scale since you are adding a constant height value. Wouldn’t dilating a binary image be quite different from dilating a gray scale image? The results from dilating the gray scale image would be dependent on the scale range of the gray scale since you are adding a constant height value.

]]> By: Steve http://blogs.mathworks.com/steve/2006/10/23/nonflat-grayscale-dilation-and-erosion/#comment-22319 Steve Wed, 11 Nov 2009 12:18:34 +0000 http://blogs.mathworks.com/steve/?p=91#comment-22319 Fredric—Use the <tt>'ball'</tt> with <tt>strel</tt>. You can also construct a structuring element of arbitrary shape by specifying the domain (nhood) and height values yourself. Fredric—Use the 'ball' with strel. You can also construct a structuring element of arbitrary shape by specifying the domain (nhood) and height values yourself.

]]> By: Fredric http://blogs.mathworks.com/steve/2006/10/23/nonflat-grayscale-dilation-and-erosion/#comment-22318 Fredric Wed, 11 Nov 2009 09:44:46 +0000 http://blogs.mathworks.com/steve/?p=91#comment-22318 Is there any hemispheric structuring element coding in MATLAB? Is there any hemispheric structuring element coding
in MATLAB?

]]> By: Steve http://blogs.mathworks.com/steve/2006/10/23/nonflat-grayscale-dilation-and-erosion/#comment-22301 Steve Fri, 06 Nov 2009 17:23:40 +0000 http://blogs.mathworks.com/steve/?p=91#comment-22301 Kezia—Try <tt>imrotate</tt>. Kezia—Try imrotate.

]]> By: kezia http://blogs.mathworks.com/steve/2006/10/23/nonflat-grayscale-dilation-and-erosion/#comment-22299 kezia Fri, 06 Nov 2009 15:56:31 +0000 http://blogs.mathworks.com/steve/?p=91#comment-22299 steve, how to perform rotation of structuring element by 15 degrees. kindly answer my question. thank u kezia joseph steve,
how to perform rotation of structuring element by 15 degrees.
kindly answer my question.
thank u
kezia joseph

]]> By: Steve http://blogs.mathworks.com/steve/2006/10/23/nonflat-grayscale-dilation-and-erosion/#comment-21757 Steve Wed, 13 May 2009 12:18:31 +0000 http://blogs.mathworks.com/steve/?p=91#comment-21757 Christina—I do not understand why you are using the HEIGHT argument at all. NHOOD controls the shape of the structuring element; make that 3-D. Your threshold value will have to be applied as a separate step; erosion does not do that. Christina—I do not understand why you are using the HEIGHT argument at all. NHOOD controls the shape of the structuring element; make that 3-D. Your threshold value will have to be applied as a separate step; erosion does not do that.

]]>