Can anyone help with this error? (I just want to see the contour of the bluish area at the bottom of the image)

close all 
clear all
clc
I = imread('frame75.jpg');
imshow(I)
imcontour(I,3)

Error: 
Error in imcontour>ParseInputs (line 110)
validateattributes(a,{'uint8','int16','uint16','double','logical','single'}, ...

Error in imcontour (line 40)
[x,y,a,extra_args] = ParseInputs(varargin{:});

Error in contour (line 8)
imcontour(I,3)

Answer 1

[darova]

Detect blue region properly

I0 = imread('frame75.jpg');
I1 = double(I0);
    % colors from blue region
R = [80 50 0];
G = [160 110 70];
B = [180 120 70];
B1 = logical(I1(:,:,1)*0);
    % detect regions and merge
for i = 1:length(R)
    B0 = abs(I1(:,:,1)-R(i)) < 20 & ...
        abs(I1(:,:,2)-G(i)) < 50 & ...
        abs(I1(:,:,3)-B(i)) < 50;
    B1 = B1 | B0;
end
% B1 = uint8( cat(3,B1,B1,B1) );
imshowpair(I0,B1)
% imshow(B1.*I0)

[muhammad choudhry]

Hi thanks for the info.

how did you detect colors from the blue region ?

R = [80 50 0];

G = [160 110 70];

B = [180 120 70];

is it possible to contour the blue region only?

[darova]

• how did you detect colors from the blue region ?

I used impixelinfo

imshow(I0)
impixelinfo

• is it possible to contour the blue region only?

Sure, try this

B2 = bwareaopen(B1,100);        % remove small regions (100 pixels)
h = strel('disk',3,0);          % dilation instrument
B3 = imdilate(B2,h);            % dilate image
B4 = edge(B3);                  % detect edge
[ii,jj] = find(B4);             % find coordinates of each edge pixel
imshow(I0)
hold on
plot(jj,ii,'.r')
hold off
zoom on

[muhammad choudhry]

Thanks alot! you being really helpful: I want few explanation from you to understand few commands mentioned above!

1) B1 = logical(I1(:,:,1)*0); % is that command means reading the first row and column of red channel in I1 then in logical command 1 means true and 0 means false.....in this command it multiplied by 0 => what does that mean?

2) In loop shown below:

or i = 1:length(R)

B0 = abs(I1(:,:,1)-R(i)) < 20 & ... %absolute value of I1-R(i)

abs(I1(:,:,2)-G(i)) < 50 & ...

abs(I1(:,:,3)-B(i)) < 50;

B1 = B1 | B0;

% what does R(i) means and why we use <20, <50.....

what does B1 = B1 | B0 means

Apologise for asking so much, as I am new to image analysis and learning while I am processing the images through this platform! your help is much appreciated.

[darova]

1) B1 = logical(I1(:,:,1)*0); % is that command means ...

No, it's just faster way to preallocate a matrix

[m,n,k] = size(I1);     % I1 3d matrix
B1 = zeros(m,n);        % B1 2d matrix (logical)

2) In loop shown below:

R(i) means red channel, G(i) means green channel, B(i) mean blue channel

You have different "blue" colors so i picked 3 of them:

R = [80 50 0];
G = [160 110 70];
B = [180 120 70];

Here is a simple example:

Imagine you have this image

i generated it using these command

[x,y,z] = peaks(50);
imagesc(z)
impixelinfo

Let's find colors in this region (value -6 approximately). How will you find them?

Let's find values

z1 = abs(z--6) < 2;         % '2' is our tolerance
imshowpair(z1,z)            % show images together

Green region is the region where values inbetween [-4 ... -8] ()

It's an example for gray image. Color image (RGB) has 3 matrices (Red, Green and Blue) so that is why i used tolerance for each channel.

Try to play with those values, see if you can detect better contour

3) what does B1 = B1 | B0 means

It means logical operand OR (simply '+'). As you can see we are searching for 3 colors (3 regions), so each for loop we are detecting different region. I want to merge them

Try to change this line

% for i = 1:length(R(i))
for i = 1

Here are results for each for loop

i = 1 i = 2 i = 3

[muhammad choudhry]

Hi darova,

Thank you for the explanation. I am nearly finish with this project. If you can help me by creating the loop which will count the pixels in the interface that will be great. I have attached the image...

Explanation: I have captured the interface of the dye by using the code we discuss above now i want to see the pixels in frame 15, 16 , 17 and see the change

[darova]

You want to calculate number of pixels inside the contour (area) or perimeter?

[muhammad choudhry]

perimeter! as perimeter explaining the rough estimation of amount of pixel in the interface and if you see in the attached cropframe15.png everything is white background except the perimeter so I want to explain in the code that colour of pixels in the perimeter are x y z hence count the pixels in the perimeter and give me the total number of pixels....

Thanks

[darova]

Number of pixels and perimeter are different things

The number of pixels but length of lines are different

Experiments with regionprops

I0 = false(100);
I0(30,30) = true;
I1 = imdilate(I0,strel('disk',10,0));
props = regionprops(I1,'perimeter');
props
2*pi*10
imshow(I1)

results

props = 
    Perimeter: 65.9411
ans =
   62.8319

[muhammad choudhry]

Hi Darova,

Sorry, I want the number of pixels in the contour (when you said perimeter I thought you mean number of pixels in boundary)..... I counted manually there are approximately 486 pixels. How would I can make a code of it to calculate them pixels.

I have attached you the image, which is basically the zoomed in image so one can count the pixels manually, but if I want to make a code for it how?

what you have explained above it is still very useful and cleared my thoughts regarding the pixels and parimeter calculations.

[darova]

binarize image

A1 = im2bw(A0);
sum(~A1(:))

[muhammad choudhry]

You the man! seriously..... It worked

I got the answer

ans = 431

just wondering when I counted manually answer was 486 and with binarizing 431... differrence in pixels are like 55. Do you know why there is that difference?

sum(~A1(:)) => means it is adding all the pixels in the image except white background

[darova]

try to change threshold

th = graythresh(I0);
I1 = im2bw(I0,th);
imshow(I1)

Answer 2

[Ameer Hamza]

You need to specify a single channel image to imcontour(). For example, just provide the blue channel

imcontour(I(:,:,3), 3)

or convert image to grayscale

imcontour(rgb2gray(I), 3)

[muhammad choudhry]

Hi,

Hamza please see the attached. I have changed to grey scale function to capture the contour of the injected dye flow but I am also getting the contours of other background is there a way to only capture the dye contour.

here is a code I used!

close all
clear all
clc
%Read the Image
I = imread('frame75.jpg');
I1 = imread('frame80.jpg');
warning('off', 'Images:initSize:adjustingMag');

%Convert to greyscale (contour command works on greyscale images)
img=rgb2gray(I); 
img1=rgb2gray(I1);

%Subtraction of the frames
figure()
dif=img-img1;
warning('off', 'Images:initSize:adjustingMag');
imshow(dif)
hold on

% contour of the image 
figure()
imcontour(dif,1);