Comments on: Pad values in dilation and erosion http://blogs.mathworks.com/steve/2007/03/23/pad-values-in-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. Fri, 10 Feb 2012 18:55:59 +0000 hourly 1 http://wordpress.org/?v=3.2.1 By: Steve http://blogs.mathworks.com/steve/2007/03/23/pad-values-in-dilation-and-erosion/#comment-21913 Steve Thu, 25 Jun 2009 11:19:34 +0000 http://blogs.mathworks.com/steve/2007/03/23/pad-values-in-dilation-and-erosion/#comment-21913 Similar to convolution, in dilation the structuring element is reflected through its center (rotated 180 degrees) before the max operation is applied. This definition of dilation is common but not universal. I believe it is more common in Europe to define dilation without the reflection. Similar to convolution, in dilation the structuring element is reflected through its center (rotated 180 degrees) before the max operation is applied. This definition of dilation is common but not universal. I believe it is more common in Europe to define dilation without the reflection.

]]> By: Jogging http://blogs.mathworks.com/steve/2007/03/23/pad-values-in-dilation-and-erosion/#comment-21912 Jogging Thu, 25 Jun 2009 03:11:13 +0000 http://blogs.mathworks.com/steve/2007/03/23/pad-values-in-dilation-and-erosion/#comment-21912 Hi, Steve I don't understand the result of imdilate (f, [1 0 0]).Take 23 of the first row, the output should be the maximum value of all the pixels in the input pixel's neighborhood under the structuring element.The neighbors are 22 23 15, and the structuring element is 1 0 0.So the neighbor used for compute maximum value should only be 22. So the output should be 22.But the matlab gives the output of 15. I can't explain it. Best Regards Jogging Hi, Steve
I don’t understand the result of imdilate
(f, [1 0 0]).Take 23 of the first row,
the output should be the maximum value of all
the pixels in the input pixel’s neighborhood
under the structuring element.The neighbors are
22 23 15, and the structuring element is
1 0 0.So the neighbor used for compute maximum
value should only be 22. So the output should be
22.But the matlab gives the output of 15.
I can’t explain it.

Best Regards
Jogging

]]> By: Steve http://blogs.mathworks.com/steve/2007/03/23/pad-values-in-dilation-and-erosion/#comment-19469 Steve Tue, 29 Jan 2008 01:41:22 +0000 http://blogs.mathworks.com/steve/2007/03/23/pad-values-in-dilation-and-erosion/#comment-19469 Eric—"Ideal" by what criterion? Every common structuring element shape has its own valid uses. <tt>strel('disk',...)</tt> is intended to be used to create a "circular" structuring element. However, by default it creates a structuring element that is only approximately circular. It does this in order to make a decomposed structuring element that is much faster to compute. Use <tt>strel('disk',R,0)</tt> to get a true circular structuring element. Just keep in mind that computing dilations and erosions with it will be slower. Use <tt>imshow(getnhood(se))</tt> to display the structuring element neighborhood as an image. Eric—”Ideal” by what criterion? Every common structuring element shape has its own valid uses. strel('disk',...) is intended to be used to create a “circular” structuring element. However, by default it creates a structuring element that is only approximately circular. It does this in order to make a decomposed structuring element that is much faster to compute. Use strel('disk',R,0) to get a true circular structuring element. Just keep in mind that computing dilations and erosions with it will be slower. Use imshow(getnhood(se)) to display the structuring element neighborhood as an image.

]]> By: Eric http://blogs.mathworks.com/steve/2007/03/23/pad-values-in-dilation-and-erosion/#comment-19458 Eric Mon, 28 Jan 2008 19:41:44 +0000 http://blogs.mathworks.com/steve/2007/03/23/pad-values-in-dilation-and-erosion/#comment-19458 I am working with medical image segmentation software specifically for calcifications in coronary arteries. Our simulations use circular calcifications that I'd like to use image segmentation to characterize. It's my understanding that the ideal structural element to use for morphology (imopen etc..) is a circular structuring element. I've noticed that MATLAB seems to be missing this feature 2007a (student). I was wondering about circular structuring elements. Is there a way to choose settings for the MATLAB disk structuring element to get a circle? Also is there a way to imshow(structureing_element) so I could visually verify it's circular. The thing is that imshow doesn't seem to like structuring elements. Thank you! -Eric I am working with medical image segmentation software specifically for calcifications in coronary arteries. Our simulations use circular calcifications that I’d like to use image segmentation to characterize. It’s my understanding that the ideal structural element to use for morphology (imopen etc..) is a circular structuring element. I’ve noticed that MATLAB seems to be missing this feature 2007a (student).

I was wondering about circular structuring elements. Is there a way to choose settings for the MATLAB disk structuring element to get a circle?

Also is there a way to imshow(structureing_element) so I could visually verify it’s circular. The thing is that imshow doesn’t seem to like structuring elements.

Thank you!

-Eric

]]>