Measuring Line Profiles in IMages

I have a image such as below which has horizontal and vertical lines in the field of view. I want to segment the vertical line adjacent to the dark area such as in red below and vertical line in the intersection of the horizontal line(as seen in black).The full image has many such structiures. I want to stack all similar segmented structures within the fov (Average measuremnet of red and black) and display.I also want to measure the average x and y as shown below if the red line concaves at the max conving location in y.

Answer 1

[Matt J]

I=im2double(im2gray(imread('image.png')));

BW=imbinarize(I);

D=bwareaopen(abs(gradient(I))>0.06,20);
D=imclose(D,strel('disk',5));

black=(median(BW,2)>0);
red=~black;

L=D.*red+2*D.*black;

imshow(labeloverlay(I,L,Colormap=[1,0,0;0,0,0])); 

●

[Vivek]

Matt, Have a follow up.How would I stack all the left segmented red edges and right edges (IN red) to understand the behaviour of the profile in the FOV.

[Matt J]

If it is a "follow-up", does that mean the original question was resolved?

Aside from that, it's not clear to me what "stack" means here. The segmented regions will all be of different sizes, so there is no natural "stacking" of them that I can comprehend. You can certainly extract each of the red regions using,

regions=regionprops(red,'Image');

You can then separate them into left and right edges by doing,

N=numel(regions);
isLeft=false(N,1);
for i=1:N
    
    tmp=mean(regions(i).Image,1);
    
    isLeft(i)=tmp(1)<tmp(end);
    
end

redLeft=red(isLeft);
redRight=red(~isLeft);

[Vivek]

What I meant is I want to overlay all the left and right edges to get a visual idea of how the profile varies

[Matt J]

But again, how will you "overlay" them when they are not all the same size? If you will crop them to a common size, what is to determine the boundaries of the cropped region?