how to get x,y pixel values of each object separately in labeled watershed image?

Hello all. I have extracted the tomato using watershed and labeled them. Now I want to know x, y, RGB, all values regarding this. I can get values for all objects in image but I want to get the values of each object in the image separately.

i want pixel values of each fruit in image separately with its color, because i want to map these values on point cloud, so that i can recognize that which fruit is which one. How I can do it? Kindly guide me as soon as you all can. Thanks.

See attached image: .

Answer 1

[Image Analyst]

watershed() gives you the labeled image, L. To get the x, y, R, G, B values, extract each label one at a time.

% Extract the individual red, green, and blue color channels.
redChannel = rgbImage(:, :, 1);
greenChannel = rgbImage(:, :, 2);
blueChannel = rgbImage(:, :, 3);
numRegions = max(L);
for k = 1 : numRegions
    thisRegion = ismember(L, k);
    [y, x] = find(thisRegion);
    redValues = redChannel(thisRegion);
    greenValues = greenChannel(thisRegion);
    blueValues = blueChannel(thisRegion);
end

[Muhammad Hammad Malik]

thanks for your kind reply, but here i am still confuse. i mentioned a piece of the code below, using this i am getting a labeled Lrgb watershed image which i posted earlier. and i want to get values of each fruit separately from this one image,like first values of fruit 1 then 2,3,4.at a time only one fruit then for next.also what is the rgbImage here?thanks

L = watershed(gradmag2);

I4 = I;

I4(imdilate(L == 0, ones(3, 3)) | bgm | fgm4) = 255;

figure

imshow(I4)

title('Markers and object boundaries superimposed on original image (I4)')

Lrgb = label2rgb(L, 'jet', 'w', 'shuffle');

figure

imshow(Lrgb)

imwrite(Lrgb,'F:\kinectdata\TomatoExtraction\tomato.png');

title('Colored watershed label matrix (Lrgb)')

[Image Analyst]

Not sure what you're doing or why you didn't use my code. I don't know why you're dilating a labeled image. And I don't know what the bgm and fgm4 images are and why you're ORing them with the dilated background image. Exactly what is I4 supposed to be???

[Muhammad Hammad Malik]

i used watershed to separate the fruits.i cannot understand about your code well.what rgbimage you are calling in code. bgm is used to create Watershed ridge lines to separate fruits and fgm4 is a bwareaopen function.in I4 marker and object boundries superimposed on original gray image.

i want pixel values of each fruit in image separately with its color, because i want to map these values on point cloud, so that i can recognize that which fruit is which one.

in your code which one i supposed as rgb image and L,where i should put your code?cannot get well.thanks

[Image Analyst]

Still not sure what effect

I4(imdilate(L == 0, ones(3, 3)) | bgm | fgm4) = 255;

will produce. It might be best if we started from the very beginning. Post your original image plus all the missing code up to the

L = watershed(gradmag2);

line.

[Muhammad Hammad Malik]

ok see the code. this is the the missing watershed code and rgb image.

"purpose of getting these pixel values is to map them with pointcloud, so that i can recognize which fruit is which one.see attach point cloud image which i get after mapping of pixel values of whole labeled watershed image but dont have the color.so i cannot recognize which one is which fruit".so i want point cloud color same as the watershed image, so that i can recognize or localize the fruit in point cloud.

rgb image=imread('F:\kinectdata\fruit');
I = rgb2gray( rgb image);
imshow(I) 
text(732,501,'Image courtesy of Corel(R)',...
     'FontSize',7,'HorizontalAlignment','right')
 hy = fspecial('sobel');
hx = hy';
Iy = imfilter(double(I), hy, 'replicate');
Ix = imfilter(double(I), hx, 'replicate');
gradmag = sqrt(Ix.^2 + Iy.^2);
figure
imshow(gradmag,[]), title('Gradient magnitude (gradmag)')
L = watershed(gradmag);
Lrgb = label2rgb(L);
figure, imshow(Lrgb), title('Watershed transform of gradient magnitude (Lrgb)')
se = strel('disk', 5);
Io = imopen(I, se);
figure
imshow(Io), title('Opening (Io)')
Ie = imerode(I, se);%
Iobr = imreconstruct(Ie, I);
figure
imshow(Iobr), title('Opening-by-reconstruction (Iobr)')
Ioc = imclose(Io, se);
figure
imshow(Ioc), title('Opening-closing (Ioc)')
Iobrd = imdilate(Iobr, se);
Iobrcbr = imreconstruct(imcomplement(Iobrd), imcomplement(Iobr));
Iobrcbr = imcomplement(Iobrcbr);
 
figure
imshow(Iobrcbr), title('Opening-closing by reconstruction (Iobrcbr)')
Iobrcbr = imhmin(Iobrcbr,80);
figure
imshow(Iobrcbr), title('Regional minima on Opening-closing by reconstruction (Iobrcbr)')

fgm = imregionalmax(Iobrcbr);
figure
imshow(fgm), title('Regional maxima of opening-closing by reconstruction apply on regional minima') %openclose
I2 = I;
I2(fgm) = 255;
figure
imshow(I2), title('Regional maxima superimposed on original image (I2)')
se2 = strel(ones(5,5));
fgm2 = imclose(fgm, se2);
fgm3 = imerode(fgm2, se2);
fgm4 = bwareaopen(fgm3, 20);%
I3 = I;
I3(fgm4) = 255;
figure
imshow(I3)
title('Modified regional maxima superimposed on original image (fgm4)')
bw = imbinarize(Iobrcbr);%
figure
imshow(bw), title('Thresholded opening-closing by reconstruction (bw)')

D = bwdist(bw);
DL = watershed(D);
bgm = DL == 0;
figure
imshow(bgm), title('Watershed ridge lines (bgm)')
gradmag2 = imimposemin(gradmag, bgm | fgm4);
L = watershed(gradmag2);
I4 = I;

[Image Analyst]

Are you saying the original image is that mostly black image with just a few red blobs, and you're somehow trying to differentiate them? Like one is a cherry, one is a tomato, and the other one is an apple? I don't think that image is good enough to do that.

And which line of the code produced the scatterplot you posted, and what does each axes represent?

[Muhammad Hammad Malik]

no.all objects in labelled watershed image are of same type. i want pixel rgb,x,y values of each object from watershed image separately.for example first get values of yellow fruit from watershed image, then blue, then red etc.so that i can map these values separately on scatter 3D plot and localize in 3D.

that is different code for point cloud generation.

now what i want is, i want values of each fruit separately from labelled watershed image with its color.

[Image Analyst]

You can find the (x,y) coordinates by using find, if you want an N by 2 list of (x,y) coordinates

[y, x] = find(binaryImage);

[Muhammad Hammad Malik]

why binary image? i want x,y,color values of each fruit separately from watershed image one by one for segmentation of point cloud.

[Image Analyst]

The binary image is used as a mask so that you can extract the colors from only that single region. Look at my code that I originally posted. thisRegion is a binary image of just one region - not all of them. Like it's ONLY region 1, then it's ONLY region 2, etc. Then find() gets you the x,y gets you the x,y values and the code below that gives you the colors for the corresponding x,y locations.

[Muhammad Hammad Malik]

so i replace thisRegion with my binary image and rgbimage is the original extracted fruit and what to do with "L" replace with what?

[Image Analyst]

You don't replace thisRegion with anything. Leave it.

You can replace L with whatever name you gave to the labeled output image when you called watershed(). How did you call watershed()? You said you called watershed, quoting: "extracted the tomato using watershed". So what does that line look like? What is the output variable from that called? I think they use L in the help/example but you could have chosen any name you want. What name did you use?

[Muhammad Hammad Malik]

start watershed with using 'L' but the resultant labelled watershed out put image is called Lrgb, and what is the 'rgbimage' here?

[Image Analyst]

rgbImage is whatever you called your original RGB image. Maybe you called it (unfortunately) "I" or maybe you called it img or some other name - I don't know what you called it.

If you put your watershed() output into a variable called "Lrgb" then use that instead of "L" of course.

A lot of times you'll see people upload code here, and of course they have to pick names for their variables. It's assumed that the poster will know to replace those names with their own names if they want them to be different.

[Muhammad Hammad Malik]

in watershed image you can see it contains four objects of different colors and i need to get x,y pixel values of each color.did you get it sir?thanks