Could anybody tell me how to replace these loops with either parfor or vectorization as the code right now runs extremely slowly?

[rows, columns, numberOfColorChannels] = size(face); % face is an image matrix

for x=1:rows

for y=1:columns

if(red(x,y)>=150)

n = n+1;

rtot = sum(double(red(:)));

end

end

end

rmean= rtot/n;

for x=1:rows

for y=1:columns

red_pixel= red(x,y);

if(red_pixel>=200)

if(red_pixel> rmean)

blush= blush+1;

rblush = sum(double(red_pixel(:)));

end

end

end

end

perc_blush(i)= (blush/(rows*columns))*100;

perc_pale(i)= ((abs(rblush-rmean))/(rows*columns))*100;

[darova]

You are just summing all pixels if any one of them is bigger than 150?

rtot = sum(double(red(:)));

[Aishwarya Jayachandran]

yes, i'm taking the sum of all the pixel values whose 'red' component is greater than 150

[Walter Roberson]

No, that is not what your code does. Every time you detect even one pixel with red component > 150, you sum(double(red(:))) which is summing all of the red for the entire image, not just the one pixel.

[Aishwarya Jayachandran]

oh okay, what i was trying to do was

as im traversing the matrix, i check if each pixel has a red value of greater than 150. and if it does, take add it to sum and do so for the whole matrix.

how do i modify the code to do this?

[Aishwarya Jayachandran]

The code i have right now is extremely slow. im going through a dataset of about 500 pictures and it takes hours for me to get the data and put it into an excel sheet. Here is the full code.

%Detect objects using Viola-Jones Algorithm

%To detect Face

FDetect = vision.CascadeObjectDetector;

%Read the input image

source=dir('C:\Users\0wner\Desktop\CBIR project\dataset\normal\*.jpg');

for i=1:566

I= imread(strcat('C:\Users\0wner\Desktop\CBIR project\dataset\normal\', source(i).name));

%Returns Bounding Box values based on number of objects

BB = step(FDetect,I);

for j=1:size(BB,1)

face= imcrop(I, BB(j,:));

end

red= face(:,:,1);

blush=0;

n=0;

rtot=0;

rblush=0;

[rows, columns, numberOfColorChannels] = size(face);

for x=1:rows

for y=1:columns

if(red(x,y)>=150)

n = n+1;

rtot = sum(double(red(:)));

end

end

end

rmean= rtot/n;

for x=1:rows

for y=1:columns

red_pixel= red(x,y);

if(red_pixel>=200)

if(red_pixel> rmean)

blush= blush+1;

rblush = sum(double(red_pixel(:)));

end

end

end

end

perc_blush(i)= (blush/(rows*columns))*100;

perc_pale(i)= ((abs(rblush-rmean))/(rows*columns))*100;

end

perc_face_blush= perc_blush';

perc_face_pale= perc_pale';

xlswrite('blush.xls', perc_face_blush);

xlswrite('pale.xls', perc_face_pale);

[Walter Roberson]

bound=max(200-1,rmean);
mean2(red_pixel>bound)*100

No looping.