Add a defined line (between two calculated points) to an image and make it a part of the image

Hello,

I have a binary image, to which I want to add a line between two points A [x1,y1] and B [x2,y2] (these points are extracted from the image using a code). I want to add this line to the image (alter the pixels of the along the line) and convert all the pixels below the line (equivalent to shading under the plot) to black.

Please note that the line to be added between the two defined points will be drawn in the white area of the binary image, therefore the line will be distinguishable. I have attached the image for your reference.

I used the following code, but it didn't work -

Im_in = imread('20170807_Test7_test_concatd_volume_CROPPEDnROT97_789x867x1001_0001.tif'); %input file
[szy, szx] = size(Im_in); % Get size of the original image
x = 100; %Define x for extending the image
z = 250; %Define z for extending the image
Im_out = uint8(255*ones(x+szy, x+szx+z)); %Define new image with increased size
singlelineimage = false(size(Im_out)); %Defining mask image for the to-be-added line
Im_out(x+1:end,x+1:x+szx) = Im_in; %Put original image in Im_out
hLine = line([1 60],[1 50]); %example line between [1,1] and [50,60]
singlelineimage = hLine.createMask();
Im_out(singlelineimage) = 0; %burn line into image by setting it to 0 or black where mask is true
imshow(Im_out);

Looking forward to your suggestions.

Thank you.

Somsubhro

[Somsubhro Chaudhuri]

Hello,

I've managed to make it work with the code below. Trying to figure out now how to change the grayscale exactly below the line from 255 to 0.

clc;
clear all;
close all;

%Read and show original image
Im_in = imread('20170807_Test7_test_concatd_volume_CROPPEDnROT97_789x867x1001_0001.tif');
subplot(2,2,1);
imshow(Im_in);
title('1: original image');

% Get size of the original image
[szy, szx] = size(Im_in); 

%Find point A (First column + Find first row to be black)
xa = 1; 
ya = (find(Im_in(:,1)==0, 1, 'first'));

%Find point B (Last column + Find first row to be black);
xb = szx; 
yb = find(Im_in(:,szx)==0, 1, 'first'); 

% Define x and z
x = 100; 
z = 250; 

%Define increased picture size
Im_out = uint8(255*ones(x+szy, x+szx+z)); 

%Put original image in new image and display
Im_out(x+1:end,x+1:x+szx) = Im_in;
subplot(2,2,2);
h_im = imshow(Im_out); %not sure why this is required, but code fails without this.
title('2: new expanded image');

   %assigning the line to a handle for the line
  hLine = imline(gca,[1 100],[1 120]);

%creating a binary mask for the line
BinarkMask = createMask(hLine,h_im);

% only displaying the binary mask for the line
subplot(2,2,3);
imshow(BinarkMask); 
title('binary mask for the line');

% Burn line into image by setting it to 0 wherever the mask is true.
Im_out(BinarkMask) = 0;
subplot(2, 2, 4);
cla;
imshow(Im_out);
title('new image with line');

[Guillaume]

I'm unclear if you have a question. It looks like you've found out how to draw the line with the createMask function. Doesn't your code work?

Personally, I find code that has an x variable used in both dimensions (as in your Im_out(x+1:end,x+1:x+szx) = Im_in) very confusing. The x is obviously not an abscissa coordinate anymore. Consider using better and longer variable names (there's no award for the shortest variable name)

[Image Analyst]

Guillaume is right. It looks like you've done it and this line of code should set the pixels to 0:

Im_out(BinarkMask) = 0;

He's also right about you making a very common novice mistake, and that is thinking (x,y) is (row, column). IT IS NOT! You need to index your arrays as (y, x) NOT (x,y) if you're using the usual nomenclature of x being horizontal and y being vertical.

[Guillaume]

The problem is not really (x,y) vs (y,x) (he/she got that right in the size query). It's the reusing of x in both dimensions. We even have x+z which I find very confusing because I think of z as the third dimension. I think it's just bad variable names.

[Image Analyst]

Yes, a better name for "x" would have been "margin" or "padding" or something else to indicate it was the amount the image would be enlarged.

Also, you should never use size like

[szy, szx] = size(Im_in); % Get size of the original image

unless you are 100% sure it's a gray scale image. It's much safer to always use

[rows, columns, numberOfColorChannels] = size(Im_in)

This will always give the correct number of columns for either gray scale or color images, while if you leave out numberOfColorChannels then columns will really be columns*numberOfColorChannels if it's a color image - not what you want. See Steve's blog http://blogs.mathworks.com/steve/2011/03/22/too-much-information-about-the-size-function/