Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.


Deo
Posts:
50
Registered:
2/13/08


Re: connected components in a grey scale image
Posted:
May 15, 2008 4:42 AM


ImageAnalyst <imageanalyst@mailinator.com> wrote in message <821dc1f78d8f4ffeb213 d9f1ee0bb355@p25g2000hsf.googlegroups.com>... > On May 14, 3:56=A0pm, "Deo " <spli...@hotmail.com> wrote: > > "us " <u...@neurol.unizh.ch> wrote in message > > > > <g0feae$mf...@fred.mathworks.com>...> "Deo ": > > > <SNIP not happy with ML labellers... > > > > > > bwlabeln just does it in multi dimensions but it still =A0 > > > > work on black and white images ( 0 and 1s)... The image > > > am working with have values between 0 and 1... > > > > > two of the (possible) solutions > > > > > % 1) use some threshold to compute a binary > > > % 2) try whether this is more helpful > > > =A0 =A0 =A0help watershed; > > > > > us > > > > I have too many intensity values to use thresholds.. and I > > tried watershed for some time ago without much luck > >  > Deo: > Are you still looking for some kind of region growing or seed fill > algorithm? You can do simple ones with thresholding. Can you explain > how exactly you define "connected" when the image is composed of > continous gray levels, because you haven't given a good explanation > for why thresholding won't work? > Regards, > ImageAnalyst
take an example an image with values like
0.009 0.009 0.009 0.001 0.8 0.007 0.001 0.9 0.006 0.9 0.001 0.7 (simple but am working with 200x400 pixel images with can have up to 50 different intensities) so I like to label the image above like 1 1 1 2 3 4 2 5 6 7 2 8 I hope that u get what I mean...



