Determine a regions average in MATLAB - image

Question

I need some help with RGB capture in a image.
I am using impixel to manualy get RGB from a picture, but i would like to create a grid of let's say 20x20 px boxes where it will automatically tell me for each box a RGB value. So in a picture lets say i have 20 boxes it will tell me 20 RGB values. Yeah an if there is 20% or more of white space that it ignores that rgb box.
Can you point me to some links or give me a general idea how to do this.
Best regards
P.S. image is just a .jpg, the background is white an in the middle there is an item.
UPDATE
This is my code for collecting RGB using impixel
px=impixel(img);
st = num2cell(px,1);
zstup = cellfun(#sum,st);
zred = size(px,1);
rez = bsxfun(#rdivide,zstup,zred);
trez=round(rez);
What I want to do is :
http://imageshack.us/photo/my-images/696/exsample.jpg/
So every box like A1, A2, and so on will return RGB value like trez in my code.
So in my code i save my trez data in a table and it is like in excell lets say 220 | 23 | 34, now if i do that to another fruit i will have
220 | 23 | 34
123 | 212| 78
and so on...
Returning to automatization, A7 and A 15 would not be good RGB canditades because they have more then 50% white area so everything that has 20% white will be ignored.
So A31 is good and the RGB value needs to be saved.
So all in all here i would have my be 6 RGB values that would have to be automatically saved like the above example.
I know how to save to table i just need help for the gathering rgb values in every box.

Answer 1

Depending on your exact needs I see two solutions:
Downscale the image using impyramid(img, 'reduce'). This gives you a smaller image consisting of average values of the original image. Then do what you did before to access single pixels. Repeat as often as necessary to get 2x2, 4x4, 8x8 or larger "boxes".
Or you could use define a box (or arbitrary shape) as a matrix of ones and zeros and use the regionprops function in order to get information about the images content depending on the fields containing ones:
roi = zeros(size(img))
roi(1:10,1:10) = 1;
r = regionprops(roi, img, 'MeanIntensity')
average = r.MeanIntensity

Answer 2

This is my complete code for automatic color grabing from pictures in folder. So the program asks you to chose a folder and after that you get a table full of information about color and roundess. I am using this code to get color from fruits that have white background . It does everything by itself. Hope it helps someone.
clear all;
clc;
uiwait(msgbox('Chose the folder where your pictures are kept. Click OK to continue..'));
% Opening the folder
folder = uigetdir(pwd);
filePattern = fullfile(folder, '*.jpg');
jpegFiles = dir(filePattern);
for k = 1:length(jpegFiles)
baseFileName = jpegFiles(k).name;
fullFileName = fullfile(folder, baseFileName);
[pathstr, name, ext] = fileparts(fullFileName);
naziv_voca=name;
%Taking RGB color
slika = imread(fullFileName);
[redovi stupci RGBboje] = size(slika);
red_ink = floor(redovi/10);
stup_ink = floor(stupci/10);
r = 1;
c = 1;
for stupac = 1 : stup_ink : stupci
for red = 1 : red_ink : redovi
red1 = red;
red2 = red1 + red_ink;
stupac1 = stupac;
stupac2 = stupac1 + stup_ink;
red2 = min(red2, redovi);
stupac2 = min(stupac2, stupci);
crveniS = slika(red1:red2, stupac1:stupac2, 1);
zeleniS = slika(red1:red2, stupac1:stupac2, 2);
plaviS = slika(red1:red2, stupac1:stupac2, 3);
crvena(r,c) = mean2(crveniS);
zelena(r,c) = mean2(zeleniS);
plava(r,c) = mean2(plaviS);
r = r + 1;
if r >redovi
r = 1;
end
end
c = c + 1;
if c >1
c = 1;
end
end
RGB=[crvena,zelena,plava];
bijela=[255 255 255];
tolerancija = 50;
rez = RGB((abs(RGB(:,1)-bijela(1)) > tolerancija) | (abs(RGB(:,2)-bijela(2)) > tolerancija),:);
trez=round(rez);
%Taking shape
pic = rgb2gray(slika);
threshold = graythresh(pic);
bw = im2bw(pic,threshold);
fbw = ones(size(bw))-imfill(ones(size(bw))-bw);
invImg = ~fbw;
f = bwlabel(invImg);
S = regionprops(f,'Area','Perimeter','centroid');
Thr=100;
S=S([S.Area]>Thr);
score = (min(sqrt([S.Area]),[S.Perimeter]/4)./(max(sqrt([S.Area]), [S.Perimeter]/4))).^2;
score=max(score);
%Inserting data into table and creating data
if exist('tablica.mat','file')
vel=size(trez,1);
for z=1:vel
s=load('tablica');
olddata=s.data;
temp=trez(z,:);
dataCell= [naziv_voca,num2cell(temp),num2cell(score)];
data=[olddata;dataCell];
save('tablica.mat','-append','data');
end
else
stupac_rgb = num2cell(trez,1);
zstupac = cellfun(#sum,stupac_rgb);
zred = size(trez,1);
rez = bsxfun(#rdivide,zstupac,zred);
trez=round(rez);
data= [naziv_voca,num2cell(trez),num2cell(score)];
save('tablica','data')
end
end
uiwait(msgbox('Your information is saved'));