Hi Bill,
you could also search the literature for medial axis transform; some
of these methods work by searching for ridges in the EDM.
For a very simple algorithm, have a look at the trueEdmHeight method
of the ImageJ built-in MaximumFinder
http://rsb.info.nih.gov/ij/source/ij/plugin/filter/MaximumFinder.javaRemove all the height calculation, just set a pixel in the output if
a ridge or maximum has been found.
If required, you could then run "skeletonize".
The disadvantage: there is no noise tolerance, so it would be
unsuited for, e.g., a medial axis transform.
Michael
________________________________________________________________
On 19 May 2009, at 02:33, Kenneth Sloan wrote:
> First, you need to find edges...and then cull isolated edges that
> don't link together into longer ridge-like features.
>
> The first hit on Googling "ridge finding image" led me to John
> Russ's very good "Image Processing Handbook".
>
> I would start there.
>
>
> On May 18, 2009, at 6:48 PM, Bill Christens-Barry wrote:
>
>> Can anyone point me to an algorithm for detecting ridges in a
>> monochrome image? I'd like to end up
>> with a binary image comprised of single-pixel width lines along
>> the ridges. I can think of several
>> ways to do it, mostly involving looping over a range of
>> thresholdis, but these methods seem very
>> inefficient.
>>
>> Thanks.
>>
>> Bill Christens-Barry
>
> --
> Kenneth Sloan
>
[hidden email]