Separating Connected Objects Effectively in Image

4 vues (au cours des 30 derniers jours)
Serg Zhelezniak
Serg Zhelezniak le 12 Nov 2017
Commenté : Image Analyst le 12 Nov 2017
Hello, I am attempting to separate several connected objects that are often only connected by one pixel (see image). I want to have all connected objects be separate from each other i.e. remove the particular pixel that's causing the issue. I have been trying to use imerode function but I keep on removing more pixels than I would like. Any advice would be greatly beneficial.

Réponse acceptée

Image Analyst
Image Analyst le 12 Nov 2017
They are separated if you use 4-connectedness. For subsequent operations, like bwconncomp() or bwlabel(), just specify 4-connectedness instead of the default 8-connectedness.

Plus de réponses (1)

Jose Marques
Jose Marques le 12 Nov 2017
Hello. Have you tried "imopen"?
Please, post your code and figures so we can help you.
  2 commentaires
Serg Zhelezniak
Serg Zhelezniak le 12 Nov 2017
im0 = image;
thresh = 190;
%Threshold Procedure
im_1a = im0 > thresh;
im_1b = bwareaopen(im_1a,5); %remove less than 5 connected components
im_1c = imerode(im_1b,strel('disk',1)); %erode with disk of radius 1
im_1d = imdilate(im_1c,strel('disk',1)); %dilate with disk of radius 1
figure
subplot(2,1,1) imshow(im_1a) title('im_1a')
subplot(2,1,2) imshow(im_1d) title('im_2d')
Performing erosion and dilation helps a little, but there are still components being connected by single pixel that should be separate (red arrow). My question is whether it is possible just to remove that particular one pixel in a manner preserves the quantity of pixel data present in im_1a.
Image Analyst
Image Analyst le 12 Nov 2017
Again, just use 4-connected like I said in my answer and you won't have the problem!

Connectez-vous pour commenter.

Catégories

En savoir plus sur Images dans Help Center et File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by