Wavelet Compression/Reconstruction and Visualization of Pulmonary X-Ray Images for Achieving of Asbestosis Infected Patients Data - Visual Computing: Scientific Visualization and Imaging Systems - page 123

Table 6.1 Absolute size of output files to be saved

Level Min. size of saved file Type of wavelet Max. size of saved file Type of wavelet

1 17.253 MB Haar 18.9 MB Sym2

2 8.627 MB Haar 9.457 MB Sym2

4 2.158 MB Haar 2.365 MB Sym2

8 540 KB Haar 592 KB Sym2

10 35 KB Haar 38 KB Sym2

14 3.9 KB Haar 4 KB Rbio1.3

Fig. 6.4 Medical fidelity of
the reconstructed images

CR ¼ 4 : 5 1048576 = 1024 350 ¼ 13 : 6 ð 6 : 9 Þ

Results obtained by proposed algorithm are presented in Table 6.1 . Haar
wavelet has minimum moments and simplest description. This is the reason for
minimum size of the output file. Problem of Haar is granular structure of the
reconstructed image.
Figure 6.4 shows fidelity of the reconstructed images versus decomposition
level. Biorthogonal, reverse biorthogonal, and Symlet wavelets do not enter the
error before ninth level of the decomposition. However, Daubechies standard
wavelets do enter error at the third level. That is the reason for excluding it from
the further consideration on the highest levels. Therefore, error is reduced. It is
imperative to understand that 0 % error does not mean that reconstructed image is
unchanged, but that medical diagnosis is unchanged. An example of reconstructed
image is in Fig. 6.5 .
An example of what to expect is in Fig. 6.6 , where original and reconstructed
images are compared. Reconstructed images are a bit smoother and blur is
introduced, but important details are still observable.