Dear all,

what is the AR_PATT_SAMPLE_NUM parameter for? It is written in a tutorial 
that it is for the resolution employed during the normalization but why 
should this be different than AR_PATT_SIZE_X and AR_PATT_SIZE Y?

are these values tunable at run time without recompiling, is there some 
functions for that?

how about a function to change the thresholding dynamically (from the 
default 130)?

I would like to use ARToolkit to do the image segmentation, pattern 
recognition but not the computation of the pattern pose by mapping the 2D 
corner to the 3D spatial coordinates. Ideally I would like ARToolkit to 
detect that this is a valid marker among its database of markers, and to 
give me the center of the pattern area, that's it. Is there a way to do 
that with the available functions or do I need to implement new ones by 
digging into the code?

Is there a distribution available that use several images of the patterns 
laid out to determine the pattern relative spatial arrangement automatically?

thanks

Yohan

Yohan, have you talked to simon about our version?  What's missing?  
I'd love to update our stuff to (for example) take the corners from the 
ARToolkit and then feed it into a better corner finder (using gradients 
in the original image) and then a better pose estimator.


On Jul 22, 2004, at 5:53 PM, Yohan Baillot (contractor) wrote:

> Dear all,
>
> what is the AR_PATT_SAMPLE_NUM parameter for? It is written in a 
> tutorial that it is for the resolution employed during the 
> normalization but why should this be different than AR_PATT_SIZE_X and 
> AR_PATT_SIZE Y?
>
> are these values tunable at run time without recompiling, is there 
> some functions for that?
>
> how about a function to change the thresholding dynamically (from the 
> default 130)?
>
> I would like to use ARToolkit to do the image segmentation, pattern 
> recognition but not the computation of the pattern pose by mapping the 
> 2D corner to the 3D spatial coordinates. Ideally I would like 
> ARToolkit to detect that this is a valid marker among its database of 
> markers, and to give me the center of the pattern area, that's it. Is 
> there a way to do that with the available functions or do I need to 
> implement new ones by digging into the code?
>
> Is there a distribution available that use several images of the 
> patterns laid out to determine the pattern relative spatial 
> arrangement automatically?
>
> thanks
>
> Yohan