[Insight-users] Circularity and Compactness Measure

[Andinet Enquobahrie]

José Santamaría López wrote:

>Perhaps, what I am going to say could be
>trivial, but may help you:
>  
>
>The radius of the largest possible circle can be
>computed by the distance between the center of
>mass of the segmented image and the furthest point
>at the segmented one.
>  
>
Just as a caveat.....

This suggestion might not work for all types of segmented regions.  For 
some geometrical shapes,
the center of mass might be outside  the segmented image (For example, 
ring or irregular and convoluated shaped objects)

-Andinet



>And, the smallest possible radious that will enclose
>it can be computed by the distance between the center of
>mass and its closest point on the segmented image.
>
>This could be an ill scheme for solving your
>requeriments, mainly when the noise is a relevant
>factor present in the images you are using. You can
>make use of a more robust mechanism, for instance,
>take a median value for each one of the rules.
>
>I don't know whether it is implemented in ITK, but
>some components of it could help you, for instance
>a Kd-Tree to compute the closest point rule.
>
>Regards.
>
>Karthik Krishnan
>  
>
>>On Sun, 2006-04-02 at 19:25 -0500, Matt Kelsey wrote:
>>    
>>
>>>Hello All,
>>>I need to measure the circularity and compactness of a 2D segmented
>>>object. To do so, I'd like to compute the radius of the largest possible
>>>circle that will fit in the shape and the smallest possible that will
>>>enclose it. Does anyone know of a built-in ITK function that will do
>>>this?
>>>      
>>>
>>The largest medial radius is the smallest possible circle that encloses
>>your segmentation.
>>
>>The filter:
>>http://www.itk.org/Insight/Doxygen/html/classitk_1_1BinaryThinningImageFilter.html
>>will find the skeleton of your 2D shape.
>>
>>The largest binary ball when centered on the skeletal pixels that has
>>all pixels inside it "on" is the largest medial radius.
>>
>>
>>
>>I have no knowledge of algorithms that compute the circumcircle of an
>>arbitrary shape, but it seems to me that you can approximate that
>>computing the largest image extents along the x/y axis by rotating it
>>from 1 through 90 degrees in steps of 1.
>>
>>-karthik
>>
>>    
>>
>>>Thanks for any ideas,
>>>Matt
>>>_______________________________________________
>>>Insight-users mailing list
>>>Insight-users at itk.org
>>>http://www.itk.org/mailman/listinfo/insight-users
>>>      
>>>
>>_______________________________________________
>>Insight-users mailing list
>>Insight-users at itk.org
>>http://www.itk.org/mailman/listinfo/insight-users
>>
>>    
>>
>
>
>_______________________________________________
>Insight-users mailing list
>Insight-users at itk.org
>http://www.itk.org/mailman/listinfo/insight-users
>
>
>  
>