|
|
| • | |
| • | |
| • | Deconvolution by adaptive inverse modelling - the TILLvisDECO approach |
| • | |
| • | |
| • | |
Hotline
|
Deconvolution and TILLvisDECO package
Download Application Note (PDF)
Deconvolution by adaptive inverse modelling – the TILLvisDECO approach
The convolution of an imaged object by the PSF is in principle reversible - a process called
deconvolution. The TILLvisDECO software (developed together with the research group of Les Loew)
(http://www.cbit.uchc.edu/) uses an adaptive inverse modelling method to deconvolve image data.
To do this, a three-dimensional finite inverse response (FIR) filter is used to adaptively model
a measured PSF (the image of a sub-resolution fluorescent bead), visualised under the same
conditions (wavelength, objective, z-step, etc.) as the specimen of interest, and which
characterises the "corruption". The output of this process is compared to a desired response
(defined as a low, medium, or high resolution virtual PSF model in TILLvisDECO) and the
difference, or error, is used to update the FIR coefficients via a least mean squares (LMS)
algorithm. Multiple iterations of this process eventually lead to a convergence of the
coefficients, which fixes the values of the FIR filter. The adaptively modelled FIR can
then be applied to the image data.
