connected component labeling
in [Matlab]
|
From: gcet joshi on 3 Aug 2010 06:56 Hi, In my work i have found edges from image then connected component labeling is applied.It works successfully.Now from connected components i want to group the components which are having maximum number of edges and want to remove small edges.
From: ImageAnalyst on 3 Aug 2010 07:34 On Aug 3, 6:56 am, "gcet joshi" <kinjalvjo...(a)yahoo.com> wrote: > Hi, > In my work i have found edges from image then connected component labeling is applied.It works successfully.Now from connected components i want to group the components which are having maximum number of edges and want to remove small edges. ------------------------------------------------ Use ismember to find blobs having perimeters of a certain length or greater. Then "group" them - whatever that means. What DOES that mean to you?
From: gcet joshi on 4 Aug 2010 01:59 ImageAnalyst <imageanalyst(a)mailinator.com> wrote in message <3a6a6b72-2ec2-4e47-87ff-22aaf75175bb(a)k19g2000yqc.googlegroups.com>... > On Aug 3, 6:56 am, "gcet joshi" <kinjalvjo...(a)yahoo.com> wrote: > > Hi, > > In my work i have found edges from image then connected component labeling is applied.It works successfully.Now from connected components i want to group the components which are having maximum number of edges and want to remove small edges. > > ------------------------------------------------ > Use ismember to find blobs having perimeters of a certain length or > greater. Then "group" them - whatever that means. What DOES that > mean to you? I want to make cluster where there are maximum edges in the result of connected component labeling.Because in the result some extra edges are also there.
From: ImageAnalyst on 4 Aug 2010 07:42 On Aug 4, 1:59 am, "gcet joshi" <kinjalvjo...(a)yahoo.com> wrote: > I want to make cluster where there are maximum edges in the result of connected component labeling.Because in the result some extra edges are also there. ----------------------------------------------------------- Are you saying that you want to identify regions of the image where there is a high concentration of blobs? (This has nothing to do with connected components labeling, except that that is a necessary step in order to run regionprops and get the centroid locations.) If so, why don't you do a morphological closing on your binary image before you label it to join some of the blobs together?
|
Pages: 1 Prev: from function to workspace Next: Comparing variables contained in MAT files |