% Heidi Thornquist
% Christy Hard

% This program will compute the Haar transform for the selected
% image. The images will be shown upon request of the user. There
% are two types of blocking employed by this method: full matrix
% and blocked matrix. There are three types of quantizing schemes
% implemented: uniform, adaptive, and statistical (Lloyd-Max).

%clear;

%------------------Choose Image-------------------------%

imnum = menu('Choose an Image','Baby','Building','Girl');
switch imnum
case 1
imfile = 'Images/baby.tif';
case 2
imfile = 'Images/building.tif';
case 3
imfile = 'Images/kgirl.tif';
otherwise
disp('Unknown image')
end

[A,map] = tiffread(imfile);
n = length(A);
n2 = log2(n);
Im = A;

prntimin = menu('Display Input Images','Yes','No');
if prntimin == 1
% Output the original image
figure(1);
makeImage(Im);
title('Original Image');
end

%----------------Compute Haar Transform on Image Matrix------------------%
%-------------------Obtain Bit Mask for Quantization---------------------%

% Obtain information from user about bit mask
lev = input('Number of quantization levels: ');
for i = 1:lev+1
lout = sprintf('How many bits for level %d: ',i);
lnum = input(lout);
if lnum > 0
q(i) = lnum - 1;
else
q(i) = 0;
end
end

% Obtain type of transform blocking
blktype = menu('Haar Transform Blocking','Full Matrix','Blocked Matrix');
switch blktype
case 1
% Haar transform for full matrix
fignum = 2;
for i = lev:-1:1
num = 2^(i+n2-lev);
Im(1:num,1:num) = haar_col(haar_row(Im(1:num,1:num)));
if prntimin == 1
% Output the images created by the Haar transform
figure(fignum);
makeImage(Im);
fignum = fignum + 1;
end
end
qtype = menu('Quantization Type','Uniform','Adapted','Statistical');
switch qtype
case 1
[M,nbits] = makeBitMask('u',n,q);
case 2
[M,nbits] = makeBitMask('a',n,q);
case 3;
[M,nbits] = makeBitMask('a',n,q);
otherwise
disp('Unknown quantization type')
end
case 2
% Haar transform for blocked matrix
fignum = 2;
lev = 3;
for k = lev:-1:1
for i = 1:8:n
for j = 1:8:n
num = 2^k;
Im(i:i+num-1,j:j+num-1) = haar_col(haar_row(Im(i:i+num-1,j:j+num-1)));
end
end
if prntimin == 1
% Output the images created by the blocked Haar transform
figure(fignum);
makeImage(Im);
fignum = fignum + 1;
end
end
[M,nbits] = makeBitMask('b',n,q(1:lev+1));
otherwise
disp('Unknown blocking type')
end

%-------------------Quantize Image Using Bit Mask Matrix--------------------%

% Normalize transformed image, quantize image, then save it to binary file.
% Then image is retrieved from the binary file and inverse quantized.

% Normalize matrix
%[Im,temp2,dtemp] = nrmlMat(Im)
[Im,div] = nrmlMat2(Im,lev);

% Quantize matrix
if qtype==3

% Use statistical quantizing method
[Im,grk,gtk] = statcom2(Im,q);
BinOut(Im,M,imnum);
Im = istatcom(Im,q,grk);
else
% Use adaptive or uniform quantizing method
Im = Quntze(Im,M);
BinOut(Im,M,imnum);
Im = IQuntze(Im,M);
end

Im = inrmlMat2(Im,lev,div);
%Im = inrmlMat(Im,temp2,dtemp);

%------------------------Haar Inverse Transform----------------------------%


switch blktype
case 1
% Haar transform for full matrix
for i = 1:lev
num = 2^(n2-lev+i);
Im(1:num,1:num) = ihaar_row(ihaar_col(Im(1:num,1:num)));
end
case 2
% Haar transform for blocked matrix
for k = 1:lev
for i = 1:8:n
for j = 1:8:n
num = 2^k;
Im(i:i+num-1,j:j+num-1) = ihaar_row(ihaar_col(Im(i:i+num-1,j:j+num-1)));
end
end
end
otherwise
disp('Unknown blocking type')
end

% Print out transformed and quantized image if requested by user
prntimout = menu('Display Output Image','Yes','No');

if prntimout == 1
% Output the image created by the Haar transform using quantization
figure(fignum);
makeImage(Im);
end

MSE = (1/n)^2*sum((A(:)-Im(:)).^2)
PSNR = 10*log10(255^2/MSE)