I could'nt find a nearest-neighbor deblurring plugin for ImageJ.
Nearest-neighbor is a way to quickly "deconvolute" z-stacks by enhancing
the slice n based on filtering the n-1 and n+1 slices and substracting
them from slice n. It's implemented in most deconvolution software as a
"quick deconvolution" option, as opposed to "complete deconvolution",
that is, mathematical reconstruction of the data using all planes. The
generalized form is called "multiple neighbors" and take into account a
user-specified number of planes around the plane calculated.
Quoting from
http://www.vaytek.com/applicationnotedetail.asp?ID=23 :
"Nearest Neighbor deconvolution is useful when the specimen can be
imaged along the optical axis and a series of images captured and stored
to disk. The resulting data set is a volume representation of the
object. The image can be deconvolved using the nearest neighbor method.
In this approach, three consecutive images are used to deconvolve the
middle image. The image on the top and bottom of the triplet can be
thought of as windows in which out-of-focus haze from all the images
above and below the processed image must pass to reach the middle image.
This technique produces excellent results when:
* The images are sampled at the proper frequency along the z axis.
For large lens NAs of 1.3 or more, sampling should be at 0.25 microns.
For a low NA of 0.7, sampling size can be 1.0 microns.
* The scanned volume is thin (50 microns - larger distance if
sample is very transmissive)"
Do anyone know a plugin that does such a task for ImageJ ? Would it be
difficult to implement ? It would be a decent alternative (although not
as precise and only qualitative) to full deconvolution which is very
slow and hard to implement in ImageJ...
Christophe Leterrier
Locked
