How to detect the edge in following image?

How would I detect the edge drew in the picture below ? First I tried to remove noises using the median filter on the image. Then I averaged the values from row 220 to 225 to remove the artifact in the middle of the image. Then, I tried to use the edge function but it does not work. I am just not able to detect the edge I drew in the attached image. How would one do this? Also attached to this post is the first and second half of the image shown in the picture. I would appreciate any help!

What I have tried so far:

load("image.mat", "image");
for r = 220:225
    for c = 1:14824
        image(r,c) = (image(219,c) + image(226,c))/2;
    end
end

filteredImage = medfilt2(image(:, 1:14824), [6,6]);

imagesc(filteredImage);
colormap("gray");
BW = edge(filteredImage, "sobel",[]);
BW = bwareaopen(BW, 20, 4);
regionOfInterest = BW(200:350, :);
[rows, columns] = find(regionOfInterest);
[C, ia] = unique(columns);
rows = rows(ia) + 200;
e = fit(C,rows, "linearinterp");
hold on;
plot(e, C, rows);

and the result

[Image Analyst]

Make it easy for us to help you. Give us a start by giving us code to read in the two .mat file, stitching them together, and running your artifact removal algorithm on the image. We can take it from there.

[Glypton]

I will in a moment but I have no idea whether one would do it that way. I have edited my post. I am sure that it is not the correct way of doing it but I have no idea how to else solve this problem. There is just one .mat file but it is too big to upload it so I split it into two files such that I can upload it.

[Image Analyst]

Something like

% Initialization Steps.
clc;    % Clear the command window.
close all;  % Close all figures (except those of imtool.)
clear;  % Erase all existing variables. Or clearvars if you want.
workspace;  % Make sure the workspace panel is showing.
format long g;
format compact;
fontSize = 20;
s1 = load('firsthalf.mat')
s2 = load('secondhalf.mat')
grayImage = [s1.firstHalf, s2.secondHalf];
subplot(2, 2, 1);
imshow(grayImage, []);
axis('on', 'square');
title('Original Image')
impixelinfo
% Get mean value of line 219 and 226
meanLine = mean([grayImage(219, :); grayImage(226,:)], 1);
% Assign that mean line to lines 220 to 225.
grayImage(220:225, :) = repmat(uint16(meanLine), 6, 1);
% Do a median filter on it.
grayImage = medfilt2(grayImage, [5,5]);
subplot(2, 2, 2);
imshow(grayImage, []);
axis('on', 'square');
title('Artifact Removed Image')
impixelinfo
subplot(2, 2, 3);
histogram(grayImage);
% Interactively and visually set a threshold on a gray scale image.
% https://www.mathworks.com/matlabcentral/fileexchange/29372-thresholding-an-image?s_tid=srchtitle
% [low, high] = threshold(180, 65535, grayImage)
% Get average vertical profile.
verticalProfile = mean(grayImage, 2);
subplot(2, 2, 3);
plot(verticalProfile, 'b-', 'LineWidth', 2)
grid on;
% Get an edge image
hSize = 15;
edgeImage = imgaussfilt(grayImage, 3) - imgaussfilt(grayImage, 7);
subplot(2, 2, 4);
imshow(imadjust(edgeImage), []);
axis('on', 'square');
title('Edge Image')
impixelinfo

[Glypton]

Sorry for the late response. Yes something like that. I will try now to somehow extract the coordinates of which the edge is composed of and try to understand what you did there. Thank you for an approach I cn use.

[Glypton]

Hey, could you tell me what the next step would be in order to find the corresponding coordinates over the ~15000 columns in order to reconstruct the edge? I am not quite getting it.

[Image Analyst]

I'd probably scan the image column by column using find to find out what row some threshold is first exceeded. If it's not exceeded, then interpolate the missing ones. Untested code:

[rows, columns, numberOfColorChannels] = size(grayImage);
threshold = 8; % whatever...
topRows = nan(1, columns);
for col = 1 : columns
    t = find(edgeImage > threshold, 1, 'first');
    if ~isempty(t)
        topRows(col) = t;
    end
end
% Interpolate missing ones
% First get valid data
strongEdge = ~isnan(topRows);
xAll = 1 : columns;
xEdge = xAll(strongEdge);
yEdge = topRows(strongEdge);
% Now linearly interpolate missing values.
yFit = interp1(xEdge, yEdge, xAll);

You might also try some morphology functions like imclose() and friends to see if you can improve the edge image and close gaps.

[Glypton]

Yes, this was also my first approach but improving the edge and set all other pixels which are not part of the edge to 0 is for me the hardest part. After that everything else becomes easy. I will try it. Thank you for your input

[Glypton]

@Image Analyst could you tell my why you used 3 and 7 for sigma here

edgeImage = imgaussfilt(grayImage, 3) - imgaussfilt(grayImage, 7);

How did you get these values?

[Image Analyst]

Trial and error. Or you could put into a nested for loop to try all combinations and see which works best.