Splitting an edge HELP!!

Hello all,

I have the following edge extracted (in yellow) that I want to split in two different parts based on the line passing through the midpoint of each row between the two edges (black line)

Anyone knows how to do it? I attach the edge matrix (full image).

Thanks a lot!

Best

Francesco

Answer 1

[Adam Drake]

Almost certainly not the most efficient way to do this, but it works.

load edge.mat

XLim = 1:550;
YLim = 300:1015;

subBUB = BUB(YLim,XLim);

y1 = 0; y2 = 716;
x1 = 0; x2 = 425;

m = (y1 - y2)/(x1 - x2);

x = XLim;
y = m * (x - x1) + y1;

top = zeros(size(subBUB));
bottom = zeros(size(subBUB));

for r = 1:size(subBUB,1)
    for c = 1:size(subBUB,2)
        if r < m * (c - x1) + y1 % less-than because of y-axis inversion on imagesc plot
            top(r,c) = subBUB(r,c);
        else
            bottom(r,c) = subBUB(r,c);
        end
    end
end

figure
subplot(1,3,1)
imagesc(subBUB)
title('Original')
line(x,y,'Color','black')
subplot(1,3,2)
imagesc(top)
title('Top')
line(x,y,'Color','black')
subplot(1,3,3) 
imagesc(bottom)
title('Bottom')
line(x,y,'Color','black')

[Francesco Pignatelli]

Hello @Adam Drake,

yes, actually it works! Now, if I want to recontruct the two sides with a spline, how can I do it?

[Adam Drake]

Reconstruct the top and bottom pieces back together or use a spline instead of a linear function to define the split?

[Francesco Pignatelli]

I mean use a spline to "smooth" the two edges once they're splitted.

Answer 2

[Adam Drake]

Your second question was an entirely different problem. I had to vectorize the image, do a spline, then recombine. Increasing stepsize will smooth your spline but reduce the number of output pixels. You could do an initial spline at a large step size and a second spline with a stepsize of 1, but I think you can handle that. This is where I say adieu.

Please Accept the Answer! Thanks!

clc, clear variables, close all
load edge.mat

XLim = 10:550;
YLim = 300:1015;

subBUB = BUB(YLim,XLim);

y1 = 0; y2 = 716;
x1 = 0; x2 = 425;

m = (y1 - y2)/(x1 - x2);

x = XLim;
y = m * (x - x1) + y1;

top = zeros(size(subBUB));
bottom = zeros(size(subBUB));

for r = 1:size(subBUB,1)
    for c = 1:size(subBUB,2)
        if r < m * (c - x1) + y1
            top(r,c) = subBUB(r,c);
        else
            bottom(r,c) = subBUB(r,c);
        end
    end
end

figure
subplot(1,3,1)
imagesc(subBUB)
title('Original')
line(x,y,'Color','black')
subplot(1,3,2)
imagesc(top)
title('Top')
line(x,y,'Color','black')
subplot(1,3,3) 
imagesc(bottom)
title('Bottom')
line(x,y,'Color','black')

% Convert img matrices to vectors
[yt,xt] = find(top);
[yb,xb] = find(bottom);


%% Spline 
stepsize = 1; 
xxt = min(xt):stepsize:max(xt);
xxb = min(xb):stepsize:max(xb);

% Spline function requires unique values
[xt_d,ut_idx,~] = unique(xt);
yt_d = yt(ut_idx);

[xb_d,ub_idx,~] = unique(xb);
yb_d = yb(ub_idx);

yyt = spline(xt_d,yt_d,xxt);
yyb = spline(xb_d,yb_d,xxb);

figure
subplot(1,2,1)
plot(xt,yt)
set(gca, 'YDir','reverse')
title('Top')
subplot(1,2,2)
plot(xb,yb)
set(gca, 'YDir','reverse')
title('Bottom')

figure
subplot(1,2,1)
plot(xxt,yyt)
set(gca, 'YDir','reverse')
title('Top')
subplot(1,2,2)
plot(xxb,yyb)
set(gca, 'YDir','reverse')
title('Bottom')

%% Recombine
% Preallocate array
out = zeros(size(subBUB));

% make x and y values integers (introduces a bit of error)
xt_out = round(xxt);
xb_out = round(xxb);

yt_out = round(yyt);
yb_out = round(yyb);

% Put 1 at x and y values
for i = 1:length(xt_out)
    r = yt_out(i);
    c = xt_out(i);
    out(r,c) = 1;
end

for i = 1:length(xb_out)
    r = yb_out(i);
    c = xb_out(i);
    out(r,c) = 1;
end


figure
imagesc(out)
title('Recombined')
line(x,y,'Color','black')

[Francesco Pignatelli]

Hej! Of course I accept it! Thanks a ton!! :D