Re: VMs: RE: oladabas again...

[Gabriel Landini <G.Landini@xxxxxxxxxx>]

On Tuesday 29 June 2004 09:15, Jon Grove wrote:
> I always pictured the character as a kind of 'mu', but it could equally
> well be an open-topped 'p', an 'r', a 'y', or an invented glyph.

Yes, it is quite puzzling...

> Regarding colour deconvolution - is that just a way of re-expressing an rgb
> colour in terms of three different base colours? If so, then it should be
> possible to simply create a 3x3 matrix to process each rgb triple and end
> up with the values expressed in terms of three new base colours (e.g. light
> blue, purple, yellow). I think the matrix would be the inverse of the
> matrix formed from the three base colours, e.g.
>
> |C1r C1g C1b|-1
> |C2r C2g C2b|
> |C3r C3g C3b|
>
> where C1,C2,C3 are the new base colours and r,g,b are the red green and
> blue components.

Yes, the algorithm is based on an orthogonal transformation + normalisation of 
the rgb data. One has to define the vectors of the colours to be separated. 
With the vms, though, this is a bit tricky as the colours have some opacity 
and the colour of the vellum is not even.

This only works for subtractive colours (i.e. light absorbing dyes: microscopy 
stained slides, combinations of filters and inks on white paper).
I wrote an ImageJ plugin to do this here:
http://www.dentistry.bham.ac.uk/landinig/software/software.html

I am planning to write a tutorial and use some of the vms as example soon.

Cheers,

Gabriel

______________________________________________________________________
To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying:
unsubscribe vms-list