How to find distance between boundary pixels and centroid?

Question

Ayesha ch 2016-5-10

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/283465-how-to-find-distance-between-boundary-pixels-and-centroid

评论： Walter Roberson 2016-5-11

Hi I'm new in matlab and I want to find the distance between the boundary pixels of each connected component and the centroid. I'm using the following algorithm.I have found centroid but dont know how to find the boundary pixels and distance. Looking forward to your reply.

for i=1:count(boundary pixels)

distance between centroid and boundary pixels

end

sum=sum/count

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

Walter Roberson 2016-5-10

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/283465-how-to-find-distance-between-boundary-pixels-and-centroid#answer_221556

在 MATLAB Online 中打开

Once you have a logical mask of ROIs corresponding to each cell, and you have filled the regions, then http://www.mathworks.com/help/images/ref/bwdistgeodesic.html

stats = regionprops(TheBinaryROI, 'Centroid', );
cents = vertcat(stats.Centroid);
mask = logical(size(TheBinaryROI));
cents_ind = sub2ind(size(mask), round(cents(:,2)), round(cents(:,1)));
mask(cents_ind) = true;
dists = bwdistgeodesic(TheBinaryROI, mask);

this will give the distances everywhere inside the ROIs. So you can proceed from there to http://www.mathworks.com/help/images/ref/bwboundaries.html use bwbounaries to trace the boundaries of the cells, and use those to select the information out of dists. Now that I think of it, you could instead use http://www.mathworks.com/help/images/ref/bwmorph.html bwmorph() with 'remove' to get just the outside edges of the region. multiply the result by the distance transform to get just the distances from the edges to respective centroids.

2 个评论
显示无隐藏无

Ayesha ch 2016-5-10

编辑：Walter Roberson 2016-5-11

在 MATLAB Online 中打开

Thank you for the help, I've tried this:

stats=regionprops(Iout,'Centroid');
cents=vertcat(stats.Centroid);
mask=logical(size(output));
cents_ind=sub2ind(size(mask),round(cents(:,2)),round(cents(:,1)));
dists=bwdistgeodesic(output,mask);
r=bwmorph(im2double(Iout),'remove');
figure, imshow(r);
for i=1:count(dists)
    d=r*dists;
end
d=d/count(dists);

but it gives me an out of range subscript error in sub2ind. kindly help

Walter Roberson 2016-5-11

在 MATLAB Online 中打开

stats=regionprops(Iout,'Centroid');
cents=vertcat(stats.Centroid);
mask = false(size(Iout));
cents_ind = sub2ind(size(mask), round(cents(:,2)), round(cents(:,1)));
mask(cents_ind) = true;
dists = bwdistgeodesic(Iout,mask);

However, I am finding that the centroids are not necessarily inside the regions. A lot depends on how you segmented into different regions: if you are able to get a clean break between regions then you probably will not do badly, but when I tried, the regions tended to run together.

请先登录，再进行评论。

Answer 2

Image Analyst 2016-5-10

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/283465-how-to-find-distance-between-boundary-pixels-and-centroid#answer_221584

You can use bwboundaries() to get a list of x,y coordinates of your perimeters. You can use bwdist() or bwdistgeodesic() to get the distance of every pixel in each blob to the closest perimeter point. Or you can use the Pythagorean theorem and the coordinates from bwboundaries if you want an "as the crow flies" distance, which in general is different than you'll get from either bwdist function..