| Developer | sonots |
|---|
| First Edition | 05/2006 |
|---|
| Last Modified | 05/2006 |
|---|
| Language | Matlab |
|---|
This page is written in both English and Japanese.
Abstract
This is a note rather than a project to use wavedec2 from wavelet toolbox.
Math or Algorithm
Refer [1]
doHaarWavelet.m
% doHaarWavelet: Apply 2-D haar wavelet for an image.
% [C, S] = doHaarWavelet(I, R) returns the approximated
% haar wavlet decomposition of the matrix I. The wavelet
% decomposition is applied until getting overall approximation
% and the # of coefficients is reduced into R percent (such as 5).
% Outputs are the approximated decomposition vector C and the
% corresponding bookkeeping matrix S.
%
% Requirements: wavelet toolbox
% See also wavedec2, waverec2
% http://dali.feld.cvut.cz/ucebna/matlab/toolbox/wavelet/wavedec2.html
%
% Author : Naotoshi Seo
% Date : May, 2006
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function [C, S] = doHaarWavelet(I, R)
% Testing
% I = imread('../recog_pie_jpg/11/04001_B_W_11_gray_01.png');
I = double(I);
if nargin < 2
R = 100;
end
% Test dwt2
% [CA,CH,CV,CD] = dwt2(I, 'haar');
% Test reconstruction
% rec = idwt2(CA,CH,CV,CD, 'haar');
% rec = uint8(rec);
% isequal(I, rec)
% imshow(rec)
% Test wavedec2
% N = 1;
% [C,S] = wavedec2(I, N, 'haar');
% A{N} = C(1:S(N,1)*S(N,2));
% A{N} = reshape(A{N}, S(N,1), S(N,2));
% isequal(CA, A{N})
% Test reconstruction
% rec = waverec2(C, S, 'haar');
% rec = uint8(rec);
% isequal(I, rec)
% imshow(rec)
N = ceil(log2(length(I)));
[C,S] = wavedec2(I, N, 'haar');
% Test reconstruction
% C(1000:end) = 0
% rec = waverec2(C, S, 'haar');
% rec = uint8(rec);
% imshow(rec)
DIM = floor(length(C)*(R*0.01));
C = C(1:DIM);
end
An experiment of data compression by haar wavelet
Source Codes
function wavedec2Test(percent)
I = imread('lena.png');
I = double(I);
R = I(:, :, 1);
G = I(:, :, 2);
B = I(:, :, 3);
N = ceil(log2(length(R)));
[RC, RS] = wavedec2(R, N, 'haar');
[GC, GS] = wavedec2(G, N, 'haar');
[BC, BS] = wavedec2(B, N, 'haar');
DIM = floor(length(RC)*(percent*0.01));
RC(DIM+1:end) = 0;
GC(DIM+1:end) = 0;
BC(DIM+1:end) = 0;
RR = waverec2(RC, RS, 'haar');
GR = waverec2(GC, GS, 'haar');
BR = waverec2(BC, BS, 'haar');
IR = cat(3, RR, GR, BR);
IR = uint8(IR);
imshow(IR);
imwrite(IR, sprintf('lenadwt%02d.png', percent), 'PNG');
end
Results
References
