Re: color image segmentation

Check out GemIdent (at www.gemident.net) - it can segment regions of
interest in color images using a training set. Plus it's free and
open source in Java. You won't have to use Matlab nor code your own
algorithm!

Heba wrote:


Image Processing Expert wrote:

how do we can calcalate the area for each object in the image(each
object has different area).......


If the color image is segmented and the result is a label
image,
the
solution looks like this:

RegionList = unique(SegmentedImage(:));

RegionList = RegionList(RegionList > 0);

NumberOfRegions = length(RegionList);

If the segmented image is binary, the number of regions can be
retrieved this way:

RegionImage = bwlabel(SegmentedImage);

RegionList = unique(RegionImage(:));

RegionList = RegionList(RegionList > 0);

NumberOfRegions = length(RegionList);

I hope this helps.

ImageAnalyst wrote:


He strongly implied that he already has the segmented image
-
he
didn't ask HOW to segment it. He was just wanting to know
that
he
should use bwlabel() to get the number of regions. He
could
follow
up
with regionprops() for more info.
Regards,
ImageAnalyst

.

Relevant Pages

  • segmentation, area, perimeter, etc
    ... I am currently working with 128*128 big color images ob objects taken ... in front of a black background. ... Now I would like to segment the image ... Are there any open source C/C++ libraries which can do these basic ...
    (sci.image.processing)
  • Re: color image segmentation
    ... If the color image is segmented and the result is a label image, ... NumberOfRegions = length; ... didn't ask HOW to segment it. ...
    (comp.soft-sys.matlab)
  • Re: tex4ht problems
    ... please try the following segment. ... .4oo java ... ... Hopfully they'll not show broken in the posting. ...
    (comp.text.tex)
Loading