a_raw = imread('potato03.jpg','jpeg');   % reads image file into 3d array a_raw(:,:,:).  The last index
                                         %  of a_raw goes from 1:3 and indexes the R, G, B color
                                         %  values of the image, respectively.
                                        
a_big = sum(a_raw,3);      % sums up RGB values to get "intensity" of each pixel.

imagesc(a_big); axis('image'); drawnow;   % see the image in false color
pause
colormap(gray); pause                     % or convert to gray scale

% The a_big array is probably too large (too many pixels) to be directly
%  useful in your model.  You'll need to sample or interpolate somehow to
%  reduce it's size.  Here are a couple of ways to do that.

% This just samples every fifth pixel
[m_big,n_big] = size(a_big);
a_smaller1 = a_big(1:5:m_big,1:5:n_big);
imagesc(a_smaller1); axis('image'); pause

% This interpolates the image to any size grid desired, say (mi x ni)
% 
mi = 32; ni = 64;
[X,Y]   = meshgrid([0:(n_big-1)]/(n_big-1),[0:(m_big-1)]/(m_big-1));
[Xi,Yi] = meshgrid([0:ni-1]/(ni-1),[0:mi-1]/(mi-1));
a_smaller2 = interp2(X,Y,a_big,Xi,Yi);
imagesc(a_smaller2); axis('image'); pause

% Finally, here's how to use the conv2 function to blur an image
% I'll make g(x,y) a gaussian for simplicity.  You'll use the g as we computed
% in class.
[xg,yg] = meshgrid([-8:8]/8,[-8:8]/8);   % generates a 17x17 grid indexing [-1,1]x[-1,1];
gfun = exp(-4*pi*(xg.^2 + yg.^2));       % gfun is a gaussian
a_new = conv2(gfun,a_smaller2);          % do the convolution
imagesc(a_new); axis('image');