How to get the centerline of a binary region

I have a binarize image below, I am trying to find the centerline of the image such as that in red on the binarize image. I did use the bskel function but get branches. I am hoping to elimated (prone or remove) all branches.

Any suggestions will be most appreciated. Any other way to do this another than using bskel will be great too.

Answer 1

[DGM]

There's this.

A = imread('arc.png')>128;

B = bwmorph(A,'skel',inf); % skeletonize
B = imfill(B,'holes'); % get rid of any tiny closed paths
B = bwmorph(B,'skel',inf); % finish skeletonizing

% remove endpoints until only two are left
while true
    e = bwmorph(B,'endpoints');
    if nnz(e)>2
        B = B & ~e;
    else
        break;
    end
end

imshow(B)

Answer 2

[KSSV]

You can get the indices of white pixels using find and to the data you can try fitting a line or quadratic equation using polyfit.

[y,x] = find(I) ;
p = polyfit(x,y,2) ;
xi = linspace(min(x),max(x)) ;
yi = polyval(p,xi) ;
plot(X,y,'.k')
hold on
plot(xi,yi,'r')

[Image Analyst]

This is exactly how I'd do it.

[Matheus Ferreira]

How to transform the line into an image like the reply above?

[Image Analyst]

@Matheus Ferreira you have the xi and yi so you just assign them (burn them into the image).

for k = 1 : length(xi)
    row = round(yi(k));
    col = round(xi(k));
    grayImage(row, col) = 255 % or 1 or whatever value you want.
end

[Matheus Ferreira]

Great, thank you! Worked perfect!

[Woj]

Hi. Sry for the stupid question, but im very very new to matlab. I have exactly the same problem, but i dont understand how do i find the white pixels?

thank you