[ImageJ-devel] ROI -> Overlay bug

[Stephan Saalfeld]

> No, the PSF is a quality of the optics. The PSF is what is scanned. The
> impulse response is what is elongated. The measurement value that the
> pixel represents is (in my simplistic understanding of signal
> processing) an integral over time of a moving PSF that is varying in
> intensity as a function of time. The location of the pixel is calculated
> from the scan mirror and stage positions.
> 

Sorry---I have been using the wrong term.  What I was referring to is
the sampling kernel, the whole function that consists of all effects
during transfer from a physical world phenomenon into a pixel sample
value.  That includes the PSF, the motion of the sensor specimen or
sensor (motion blur), quantization of the signal in the sensor, noise
and what ever.  In a microscope, this kernel is some difficult to obtain
n-dimensional thing (space, time, wavelength, m properties of the dye
and specimen), target of the ideal deconvolution framework.

> > However, that PSF would still be symmetric, except
> > if the sensor has a tendency to collect photons only later or earlier
> > during exposure.
> 
> Such as the situation if the scan is accelerating or decelerating during
> acquisition which happens in many designs: one 'side' of the pixel will
> tend to be brighter than the other 'side' (one end of the integral will
> be longer than the other).
> 

Good to know---has it been measured ever?  Would be nice to include it
as a trivial to understand part of the above mentioned kernel.

> >  So to make my statement clearer, a point sample taken
> > from the physical world is almost never a point-sample, but it's
> > representative coordinate is most likely not at the top left corner of
> > the area between samples.
> 
> Agreed, if we want to relate pixels back to the real world in some
> reliable fashion (that's the general goal, right?)
> 

Yes.

Cheers,
Stephan