function [model] = learnDistribution(image,manual,method)

model.h=[0 0];
model.center = [0 0];
width = 0;
height = 0;

sz = size(manual);

if(sz(2)==1)
figure(1);
imshow(image); colormap(gray); drawnow;
rectangle = ginput
topLeft = rectangle(1,:);
bottomRight = rectangle(2,:);
dimensions = bottomRight - topLeft;
width = round(dimensions(1,1))
height = round(dimensions(1,2))
xoffset = round(topLeft(1,1))
yoffset = round(topLeft(1,2))
model.h = [round(width/2) round(height/2)];
model.center = [(xoffset + model.h(1)) (yoffset + model.h(2))]; 
region = image(yoffset:yoffset+height-1,xoffset:xoffset+width-1);
%figure(2);
%imagesc(region); colormap(gray); drawnow;
end

if(sz(2) == 4)
  model.center = [manual(1) manual(2)];
  model.h = [manual(3) manual(4)];
end


region = image;

%Set up the smoothing filter
smooth = [.0625 .125 .0625; .125 .25 .125; .0625 .125 .0625];

%Set up the differentiator

%Use simple difference
if(method==0)
fx=[1,-1];
fy=fx';
end

%Use first derivative of the Gaussian - i.e. Canny Edge-Detector
if(method==1)
end

imageDx = conv2(region,fx);
imageDy = conv2(region,fy);

model = defineModel(imageDx,imageDy,model,16);
illustrateDist(image,model.h,model.center,model.modelDist)

%figure(1);
%imagesc(uint8(model.modelDist));
%figure(2);
%imagesc(uint8(distribution/C));