Fractal_Box_Counter

[André Tchiftchibachian]

Dear colleagues,

I am interested by fractal analysis of a turbulent jet contour, ( cf.
Sreenivasan and Meneveau, Yale University).
Could anyone conservant with the plugin Fractal_Box_Counter, tell me if
it concerns hollow contour or if it may be used for full inside countour
(given by my digital camera)? (I am ignorant of Java...shame on me!)
In the first case, could it exist a mean to substract the jet inside,
keeping only the border ?
Later I intend to study the primary atomization drops near the contour...

Thanks a lot,

Dr A.Tchiftchibachian (F-Aix-en-Provence)

[Gabriel Landini]

On Thursday 24 August 2006 10:03, André Tchiftchibachian wrote:
> Could anyone conservant with the plugin Fractal_Box_Counter, tell me if
> it concerns hollow contour or if it may be used for full inside countour

The box counting can analyse the contour (surface fractal) or the whole set
(mass fractal). It depends what you submit to the procedure.

You can test this easily by analysing a straight line or hollow
rectangle/circle (which should be near D~1.0 and a filled image D~2.0).

Which one is relevant? It depends on the data, not the algorithm. It may be
one (surface or mass) or both or none. There is no strict relation between
the surface and the mass fractal dimension, they are different things, but
sometimes they may coincide. It is therefore important to specify which one
is being estimated.

Be aware that to estimate the surface dimension of a 3D object, you need to
analyse a 2D cut through the object (the so-called zeroset), not a projected
outline.

I hope it helps.

Gabriel

[André Tchiftchibachian]

Hi Gabriel,

Thank you for your help. Even more, your last remark raised up some  
interrogation...
I was intending to make some approximation concerning the projection and
the longitudinal cut: a kind of assumption.
First, I go to implement some trials..
So long

André

Gabriel Landini wrote:

>On Thursday 24 August 2006 10:03, André Tchiftchibachian wrote:
>  
>
>>Could anyone conservant with the plugin Fractal_Box_Counter, tell me if
>>it concerns hollow contour or if it may be used for full inside countour
>>    
>>
>
>The box counting can analyse the contour (surface fractal) or the whole set
>(mass fractal). It depends what you submit to the procedure.
>
>You can test this easily by analysing a straight line or hollow
>rectangle/circle (which should be near D~1.0 and a filled image D~2.0).
>
>Which one is relevant? It depends on the data, not the algorithm. It may be
>one (surface or mass) or both or none. There is no strict relation between
>the surface and the mass fractal dimension, they are different things, but
>sometimes they may coincide. It is therefore important to specify which one
>is being estimated.
>
>Be aware that to estimate the surface dimension of a 3D object, you need to
>analyse a 2D cut through the object (the so-called zeroset), not a projected
>outline.
>
>I hope it helps.
>
>Gabriel
>
>  
>

[Gabriel Landini]

On Thursday 24 August 2006 13:04, André Tchiftchibachian wrote:
> Thank you for your help. Even more, your last remark raised up some
> interrogation...
> I was intending to make some approximation concerning the projection and
> the longitudinal cut: a kind of assumption.

Well, think of a ball with lots of holes/tunnels/cracks. The projection may
look smooth, but the section will not.

I forgot to mention that you should use a recent version (>=1.37i) of the
ij.jar because now the built in fractal box counting takes into consideration
a "background" setting. (Before it used the smallest fraction of foreground
vs background to decide what was the object and it was not possible to
analyse objects that had more foreground than background).

Cheers,

Gabriel