Image Processing Basics - Digital Signal Processing: Fundamentals and Applications - page 742

FIGURE 14.42
(a) Wavelet coefficients; (b) original image; (c) 4:1 compression.
Figure 14.43 (c), we can observe a noticeable degradation of image quality. Since the high-frequency
details are discarded, the compressed image shows a significant smoothing effect. In addition, there are
many advanced methods to quantize and compress the wavelet coefficients. Of these compression
techniques, the embedded zerotree wavelet (EZW) method is the most efficient one; it can be found in
Li and Drew (2004).
Program 14.4. One-level wavelet transform and compression.
close all; clear all; clc
X ¼ imread( ' cruise ' , ' JPEG ' );
Y ¼ rgb2gray(X); % Convert the image into grayscale
h0 ¼ [0.054415842243144 0.312871590914520 0.675630736297712 .
0.585354683654425 -0.015829105256675 -0.284015542962009 .
0.000472484573805 0.128747426620538 -0.017369301001845 .
-0.044088253930837 0.013981027917411 0.008746094047413 .
-0.004870352993456 -0.000391740373377 0.000675449406451 .
-0.000117476784125];
M ¼ length(h0);
h1(1:2:M-1) ¼ h0(M:-2:2);h1(2:2:M) ¼ -h0(M-1:-2:1);% Obtain QMF highpass filter
[m n] ¼ size(Y);
% Level-1 transform
[m n] ¼ size(Y);