Easiest way to find the mean of a group of pixels in an image

I am attempting to analyze an image of a scanned thin section where I have isolated crystals of a certain chemical composition. My goal is to define groups of pixels at the cores of these crystals in order to find the mean value of them. Using the mean values I hope to create a frequency distribution so that I can define crystal populations based on the groupings in the distribution.

So far, I have been struggling with the "inpolygon" function to define groups of pixels. My attempts have gone along the lines of:

>a = imread('image.jpg');

>imagesc(a) %Here I zoom into my area of interest (the crystal core)

>aa = xlim; %Gives me [xmin,xmax]

>ab = ylim; %Gives me [ymin,ymax]

>ac = inpolygon(xmin,ymin,xmax,ymax)

This is as far as I've gotten. "ac" has been returning as 0, and I don't know where to go from here. Any advice would be greatly appreciated.

[Apatschinn]

UPDATE:

I have reviewed the input arguments definitions (<http://www.mathworks.com/help/matlab/ref/inpolygon.html#zmw57dd0e373179>) and I have realized that I am totally off. So I made another attempt.

>ac = inpolygon(aa,ab,aa,ab);

I tried defining it this way because my query points are the same as my polygon points. I got a 1x4 vector of ones (yay, locicals).

So now do I just define a matrix with the xlim,ylim points and then just find the mean of it?

Answer 1

[Chad Greene]

Try this:

% Here's an image: 
Z = double(imread('tire.tif'));

% It has these corresponding rows and columns: 
[cols,rows] = meshgrid(1:size(Z,2),1:size(Z,1)); 

% Perhaps you're interested in all the stuff in this polygon: 
coli = [50 53 120 170]; 
rowi = [40 140 141 60]; 

% These are the indices of all the values inside the polygon:  
ind = inpolygon(cols,rows,coli,rowi); 

% Get the mean of the Z values inside the polygon like this: 
mean(Z(ind))

image(Z)
colormap(gray)
hold on
plot(coli,rowi,'go-')

% Or you could set all the Z values inside the polygon to zero: 
Z(ind) = 0;