How do I find the centerlines shown in the image?

Original image: "myimage"

Binary image: "binary"

Code to reach from original image to binary image:

I = imread("myimage.jpg");
I = im2gray(I);
I = imadjust(I);
Igs = I > 150;
imshow(Igs)

Igs = imfill(Igs,"holes");
Igs = bwareaopen(Igs,50);
imshow(Igs);

SE = strel("disk",10);
Igs = imclose(Igs,SE);
Igs = imopen(Igs,SE);
imshow(Igs)

I want to detect the 6 centerlines shown in the binary image marked in red. What functions or method do I use to detect these centerlines?

Thankyou.

Answer 1

[Prateek]

Hi Husain,

There can be different approaches to find and draw centerlines, such as scanning image columns or using the function “bwskel”. The following links contain detailed discussions on the same:

1. Finding center of a line - MATLAB Answers - MATLAB Central (mathworks.com)
2. How to find the centreline of two curves - MATLAB Answers - MATLAB Central (mathworks.com)
3. Centerline and bounding curve in image - MATLAB Answers - MATLAB Central (mathworks.com)

Please refer to these discussions. These should be helpful in constructing a solution.

Regards,

Prateek

[Husain Signalwala]

Thankyou for your reply. I could not construct a solution from the above methods. The bwskel() operation does not give me accurate results.

Here is my code using the bwskel() operation.

I = imread("myimage.jpg");

Error using imread>get_full_filename
File "myimage.jpg" does not exist.

Error in imread (line 372)
fullname = get_full_filename(filename);

I = im2gray(I);
I = imadjust(I);
Igs = I > 150;
imshow(Igs)

Igs = bwareaopen(Igs,30);
imshow(Igs)
avg = fspecial("average",[45 45]); %average filtering
Iavg = imfilter(Igs,avg);
If = medfilt2(Iavg,[45 45]); %median filtering
imshow(If)

S = bwskel(If);
imshow(S)

[H,T,R] = hough(S); %applying the hough transform

houghMatViz(H,T,R)

peaks = houghpeaks(H,8)

lines = houghlines(S,T,R,peaks)

plotHoughLines(I,lines)


function houghMatViz(H,T,R)
    % Convert the hough matrix to a grayscale image
    Hgray = mat2gray(H);
    % Enhance its brightness
    Hgray = imadjust(Hgray);
    % Display the hough transform
    imagesc(Hgray,"XData",T,"YData",R) 
    % Assign a colormap for the image
    colormap(hot)
    title("Hough Transform")
    % Add x-y labels
    xlabel("\theta")
    ylabel("\rho")
end

function plotHoughLines(image, lines)
    % Show the image
    imshow(image)
    % Plot the hough transform lines
    hold on    
    numLines = length(lines);
    for k = 1:numLines
        xy = [lines(k).point1; lines(k).point2];
        plot(xy(:,1),xy(:,2),"LineWidth",2,"Color","red",...
            "Marker", "x", "MarkerEdgeColor", "g")
    end
    hold off
end

The "filt.jpg" shows the filtered image. The "skel.jpg" shows the image after applying bwskel(). The "hough.jpg" shows the lines detected by applying hough transform.

Is there any other way to do it or any improvements to my code that works to detect the lines?

[Prateek]

Hi Husain,

I believe you would have found a solution by now. If that is not the case, I suggest you approach the MathWorks Technical Support for this. Here's the link for it - https://in.mathworks.com/support/contact_us.html?requestedDomain=

Regards,

Prateek