local equalization

[Matteo]

I read an interesting example on John C Ross' Image Processing Cookbook.He's using a running window to perform local equalization of an image containing a fingerprint on a colored magazine cover, with a portion of fingerprint on very dark background and a portion on very light background. He rightfully claim that non-linear adjustment using curves or levels won't enhance contrast in both areas at the same time. Unfortunately the book lacks technical detaisl or reference.

I found this Photoshop Plugin online but agian it lacks details.
Can anyone recommend some reference and/or code from the exchange?

http://www.mehdiplugins.com/english/localequalization.htmThank you

[ImageAnalyst]

Matteo :
This is commonly called "contrast limited adaptive histogram
equalization" (CLAHE) and is performed by the MATLAB function
adapthisteq() in the Image Processing Toolbox. It's essentially
histogram equalization (http://en.wikipedia.org/wiki/
Histogram_equalization) but done within a sliding window so that the
type of equalization varies at each window position (because the image
varies at each window position). There are at least three ways to do
histogram equalization that I know of that give different accuracies
as compared to the theoretically flat output histogram (cdf, adding
noise, and sorting in that order but I don't want to get into their
details now). When done on an entire image it can give an okay result
- better than histogram equalizing the entire image which is almost
always not an improvement.

This would be a somewhat effective way to enhance fingerprints but far
from the best. Diffusion methods are better, for example:
http://academic.research.microsoft.com/Paper/2440625.aspx
ImageAnalyst

[Matteo]

Awesome, thank you!

ImageAnalyst <imagea...@mailinator.com> wrote in message <b305f927-4df8-4bf8...@f25g2000yqc.googlegroups.com>...