Re: correction : ts calibration

On Fri, 2004-04-16 at 14:47, Alain2210 wrote:
> Selon Koen Kooi <koen_at_vestingbar.nl>:
> >
> > > that is
> > > ycal = ax + by + c
> > > ycal = dx + ey + f
> > With this kind of formulas, isn't it "easier" to make a 240x320
> > transformation matrix and put that in tslib? Tslib can derive the
> > equations from that, or directly apply it to the readings. The matrix
> > may be smaller at the sacrifice of precision.
>
> It could be, using a calibration with more than 5 points. I do not know how it works
> with the other ipaq. Is the touchcreen acute ? Moreover I do not know how PPC works
> with this.
The matrix can also be used the other way around: generate it from the 5
calibration points. Random point (with a mininum distance of course) can
be used to calibrate it further. I will look into such a system next
week or so (I'm away this weekend and have some deadlines next week). If
you have access to a program like Maple or Matlab you could test some
stuff.

regards,

Koen
Received on Fri Apr 16 2004 - 12:59:43 EDT