removing background of the color picture

Hi, i am working on a project, were I have to calculate the pixels inside the object only, my image contains a human face and a white background, all I want to do is to remove the background and save only the face of the person and calculate histograms from the face pixels only for my project. I have tried a thresholding method, so that kind of removes background and gives a black color as background. And i have used imwrite function to save the image and use it for further analysis. but the problem, I am not sure whether I am calculating only from the face pixels or the entire image including black as background. I am attaching my code here. kindly help me clear my doubt

%%threshold the image
I = imread('facergbimg.jpg');
imshow(I)
[BW rgb]= createMaskbcgremoval(I); this is a bckground removal code done with the use of image proccessing color threshold from the math works. 
figure,imshow(rgb)
imwrite(rgb,'file.jpg')
% % save the image in a folder 
% %  work with the background removal image working only with face pixels 
% %  separate image into 3 channels 
img = imread('file.jpg'); % Read image
red = img(:,:,1); % Red channel
green = img(:,:,2); % Green channel
blue = img(:,:,3); % Blue channel
a = zeros(size(img, 1), size(img, 2));
just_red = cat(3, red, a, a);
just_green = cat(3, a, green, a);
just_blue = cat(3, a, a, blue);
back_to_original_img = cat(3, red, green, blue);
figure, imshow(img), title('Original image')
figure, imshow(just_red), title('Red channel'),imwrite(just_red,'just_red.jpg')
figure, imshow(just_green), title('Green channel'),imwrite(just_green,'just_green.jpg')
figure, imshow(just_blue), title('Blue channel'),imwrite(just_blue,'just_blue.jpg')
%% now split the histogram 
G =imread('just_green.jpg');
 K_median_cal= reshape(G,[],1);
 total_median = mean(K_median_cal);
 K(G<total_median) = G(G<total_median);
P(G>total_median) = G(G>total_median);
K = uint8(double(total_median) .* ones(size(G)));
P = uint8(double(total_median) .* ones(size(G)));
for i=1:length(G(:,1))
    for j=1:length(G(1,:))
        if G(i,j)<total_median
            K(i,j)=G(i,j);
        else
            P(i,j)=G(i,j);
        end
    end
end
figure(1)
imshow(G)
figure(2)
imshow(K)
title('lowerrectified');
figure(3)
imshow(P)
title('upper rectified');
figure(4)
imhist(G)
figure(5)
imhist(K)
ylim([0 10000])
figure(6)
imhist(P)
ylim([0 10000])

Answer 1

[Image Analyst]

"whether I am calculating only from the face pixels or the entire image including black as background." <== you are calculating total_median from G, which is the green channel of the background-removed image that you wrote to disk. I must say though that it looks like a very roundabout way of coding, like why write the rgb matrix out to disk only to read it back in as img? Etc.

[Malini Bakthavatchalam]

Hello, Image Analyst,

I learnt things from this question and answer session, so basically i had a color image, so i want to remove background and work only on the face pixels. Next I need to separate the color image into 3 channels, and from the 3 channels. I have to calculate median of the three channels. My task is to make the median histogram of the channels and split the histogram into two halfs. I want to see how the transmittion is happening from black to each color channel. since i learnt from all the online answer sessions. my code is round about. Could you help me with simple code?

*[total_median = median(K_median_cal)]. this line should be median

thanks

[Image Analyst]

Not sure what you want but to compute the median of each color channel, you'd do this:

medianR = median(red);
medianG = median(green);
medianB = median(blue);

I can't figure out how you're computing total_median or what it even means.

And I don't know exactly what "see how the transmittion is happening from black to each color channel" means. Do you want code to zoom in to see the black-to-colored region in more detail? If so, see the attached zooming demo.

[Malini Bakthavatchalam]

hi professor,

I think I wasnt clear enough , all i want to do, for example take a green channel find the median (for ex if the median of the image histogram is 214, i need to split the histogram into half below 214 and above 214 and create the image with the values less than 214(lower rectified image) and above 214(upper rectified image). i want to know how the less than median and above the median value image looks like ..

so for that this formula is the optimization formula .

K(G<total_median) = G(G<total_median);- create image with this

P(G>total_median) = G(G>total_median);& create an image with this. so by the end of it i should have three images one is the original and other two are lower and higher rectified images. this is for an color illusion study.