Re: 3D caliper

Posted by Thomas Boudier on Jan 18, 2010; 9:44am
URL: http://imagej.273.s1.nabble.com/3D-caliper-tp3689657p3689662.html

Hi,

I have implemented the max feret distance for 3D objects (by brute force
;-) ), check the roimanager3d in :
http://imagejdocu.tudor.lu/doku.php?id=plugin:morphology:3d_binary_morphological_filters:start

use first 3d-oc to label your objects and then use the "add image"
command to add your labeled objects to the 3dmanager, then 3d measure.

a new version is coming soon, with the 3d mas radial distance, the
maximum distance between the center and the border.

Thomas


Michael Schmid a écrit :

> Hi,
>
> my 3 cents (though I am clearly not an expert on this):
>
> One way for an aproximation of maximum Feret might be an approach similar
> to that used by ImageJ for minimum Feret: for a list of spatial
> directions, loop over all points of the convex hull and calculate the
> projection onto the direction.
> Get the maximum and minimum of these projections values for all points;
> the difference is the Feret diameter in that direction.
>
> If you want 1% accuracy, no direction must be more than arccos(0.99)=0.14
> radians from a direction in the list of directions. I guess that a list of
> about 150 directions should be enough for this. If your convex hull has
> more than a few hundred points, this method should be faster than the
> brute force method with a nested list - but it is an approximation...
>
> Michael
> ___________________________________________________________________
>
> On Sat, January 16, 2010 17:31, Gabriel Landini wrote:
>> On Friday 15 January 2010, Mike Doube wrote:
>>> How might one find a bounding box that has the least volume, or shortest
>>>  sum of edge lengths, or a non-brute-force Feret?
>> This seems one way for the bounding box:
>> http://www.mlahanas.de/CompGeom/opt_bbox.htm
>>
>> Here is a presentation with various methods:
>> http://www.csam.ucl.ac.be/news/slides/gorissen09.pdf
>>
>> Not sure what is the best method to find the maximum Feret, but if one is
>> tempted to do it brute force, one should be able to save a lot of time by
>> computing the convex hull first, as the max Feret *has* to be in the
>> convex
>> hull (if I am not wrong). And no, I do not think brute force is a good
>> idea.
>>
>> Cheers,
>>
>> Gabriel
>>
>
>

--
   /**********************************************************/
      Thomas Boudier, MCU Université Pierre et Marie Curie,
      IFR 83. Bat B 7ème étage, porte 706D, Jussieu.
      Tel : [phone]  Fax : [phone]
/*******************************************************/