How do we know what are the superpixels connected to each other ?

Question

Ad 2017-4-23

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/336813-how-do-we-know-what-are-the-superpixels-connected-to-each-other

编辑： Casio Uploader 2018-6-9

I am using SLIC superpixel algorithm. How to create adjacency matrix for superpixels? How do we know what are the superpixels connected to each other?

Thanks in advance

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Image Analyst 2017-4-23

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/336813-how-do-we-know-what-are-the-superpixels-connected-to-each-other#answer_264107

glcm_demo.m

You can use GLCM - the Gray Level Cooccurrence Matrix, done by graycomatrix(). It tells you which gray levels are next to which other gray levels. I attached a demo.

2 个评论
显示无隐藏无

Ad 2017-4-23

在 MATLAB Online 中打开

I have used GLCM. I got result like this.

graystats.Correlation,0.056197194015968

graystats.Energy,0.985076430158353

graystats.Homogeneity,0.245757011433993;

graystats.Contrast,0.992415460185401

 Q. How do I know the connected superpixels?
 GLCMS2= graycomatrix(rgbtogray);% it is accepting 2D images only
  graystats = graycoprops(GLCMS2,{'all'});

Image Analyst 2017-4-23

You don't need the stats. Just look at the matrix and see what locations have non-zero pixel pair counts. It says what label is next to what other label.

2 is next to 3, 4, 5, and 6

3 is next to 2, 4, 5, 6, and 7

4 is next to 2, 3, 5, and 6

and so on.

请先登录，再进行评论。

Answer 2

Casio Uploader 2018-6-9

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/336813-how-do-we-know-what-are-the-superpixels-connected-to-each-other#answer_323938

编辑：Casio Uploader 2018-6-9

glcms=graycomatrix(L); % 'L' is Labled Image output of SLIC superpixel algorithm

kkkk=glcms(:,SupNum); %if you want to find neibours of 1,then input SupNum=1 in both places

[rrrr,~]=find(kkkk>0); aa=find(rrrr==SupNum);

rrrr(aa)=[];

%%rrrr is output matrix which contains the neighbours of 1(or your desired one)

N.B.:It will give you just right directions neighbour information,to get all direction use 'Offset' optiom for graycomatrix() function.See: https://in.mathworks.com/help/images/ref/graycomatrix.html