MATLAB: How to vectorize this function

image processingmatrixvectorvectorization

Hi.

I have a series of images all with the same background but varying foregrounds. I'm looping through all these images and taking median pixel values to calculate the background so I can perform background subtraction. It works well but the code is slow because it contains a double for loop. How can I vectorize the following code?

    for x = 1:480
        for y = 1:640
            median_bg_RedValues(x,y) = median(bg_RedValues(x,y,1:actualNumOfImages));
            median_bg_GreenValues(x,y) = median(bg_GreenValues(x,y,1:actualNumOfImages));
            median_bg_BlueValues(x,y) = median(bg_BlueValues(x,y,1:actualNumOfImages));
        end
    end

median_bg_Red/Green/BlueValues is a 480*640 matrix, and bg_Red/Green/BlueValues is a 480x640x40 image, and I want to take the median of each cell's 40 elements (these 40 elements are the different pixel values for the 40 images).

How can I make this faster using vectorization?

Thanks.

Best Answer

median_bg_RedValues   = median(bg_RedValues,  3);
median_bg_GreenValues = median(bg_GreenValues,3);
median_bg_BlueValues  = median(bg_BlueValues, 3);

Related Solutions

MATLAB: Outputting arrays from function in for loop

Going to use nio.sigma instead of sigma, since sigma is a matlab function and NOT a constant that you want in your function dT.

nio.sigma     = 5.67e-8;     % Stefan Boltzmann constant

To fix the plotting, do something like this:

Data = cell(2, 3); %row1 = t1,t2,t3, row2 = T1,T2,T3
for j = 1:3
    [Data{1, j}, Data{2, j}] = dT(nio,h(j));
end
plot(Data{:})

Found some issues with your function, see comments. Make sure this is the same equation though!

function [t,T] = dT(nio,h)
%RECHECK CHANGES to make they are the same eqn!
%dT = @(t,T) ( -h*nio.A*(T-nio.Tinf) - nio.A*sigma*nio.eps*(T^4 - nio.Tinf^4)) / (nio.rho*nio.V*nio.c);
%- dont' name dT function handle same as your dT function name. It gets confusing.
%  > using aT instead
%- sigma is matlab function, so you can't use that as a variable.
%  > using nio.sigma instead
aT = @(t,T) ( -h*nio.A.*(T-nio.Tinf) - nio.A*nio.sigma*nio.eps*(T.^4 - nio.Tinf^4)) / (nio.rho*nio.V*nio.c);
[t, T] = ode45(aT, [0,60] , [nio.Ti]);

MATLAB: Change color in image

Why don't you have access to the original image? If you don't, for some strange reason, then you can find the pixels by first extracting the individual color planes, then median filtering them and replacing the colored pixels with the median filtered pixels. It will work a lot better if you have the original uncompressed image, rather than the blurry one you uploaded, which does not have sharp graphics.

rgbImage = imread(fullFileName);
% Get the dimensions of the image.  numberOfColorBands should be = 3.
[rows columns numberOfColorBands] = size(rgbImage);
% Display the original color image.
subplot(2, 2, 1);
imshow(rgbImage);
axis on;
title('Original Color Image', 'FontSize', fontSize);
% Enlarge figure to full screen.
set(gcf, 'units','normalized','outerposition',[0 0 1 1]);
% Extract the individual red, green, and blue color channels.
redChannel = rgbImage(:, :, 1);
greenChannel = rgbImage(:, :, 2);
blueChannel = rgbImage(:, :, 3);
% Then do a median filter on each one.  
medianRed = medfilt2(redChannel, [7 7]);
medianGreen = medfilt2(greenChannel, [7 7]);
medianBlue = medfilt2(blueChannel, [7 7]);
% Display the image.


subplot(2, 2, 2);
imshow(medianRed);
title('Red Median Image', 'FontSize', fontSize);
% Then find the pure color pixels.
thresholdValue = 40;
pureRed = redChannel >= thresholdValue & greenChannel < thresholdValue & blueChannel < thresholdValue;
pureGreen = redChannel < thresholdValue & greenChannel >= thresholdValue & blueChannel < thresholdValue;
pureBlue = redChannel < thresholdValue & greenChannel < thresholdValue & blueChannel >= thresholdValue;
% Display the image.
subplot(2, 2, 3);
imshow(pureRed);
title('Red Mask Image', 'FontSize', fontSize);
% Then replace each pure color with the median from that channel.
redChannel(pureRed) = medianRed(pureRed);
greenChannel(pureGreen ) = medianGreen(pureGreen );
blueChannel (pureBlue ) = medianBlue(pureBlue );
% Construct the new color image
rgbImageFixed = cat(3, redChannel, greenChannel, blueChannel);
% Display the image.
subplot(2, 2, 4);
imshow(rgbImageFixed);
title('Fixed RGB Image', 'FontSize', fontSize);

Best Answer

Related Solutions

MATLAB: Outputting arrays from function in for loop

MATLAB: Change color in image

Related Question