** Next:** Synthetic 3-D data
** Up:** APPLICATION OF WAVELET-BASED COMPRESSION
** Previous:** APPLICATION OF WAVELET-BASED COMPRESSION

The synthetic shot gather consists of three hyperbolas.
The dataset has been contaminated by random noise. For this 2-D dataset,
*n*_{1}=512 and *n*_{2}=128. Figure 3,
4, 5, and 6 are the
*t-x* domain and *f-k* domain results.

**compcsg
**

Figure 3 Synthetic shot gather in the
*t-x* domain (*scale* = 0.715). In this case, the compression ratio is 50 and SNR is 7.69dB. The two figures are plotted on the same scale.

**a**: Original dataset.

**b**: Compressed/decompressed dataset.

**compcsgdiff
**

Figure 4 Difference of the two results in last figure. There is little remnant of the three hyperbolas in the figure.

**a**: Difference shown on the absolute scale.

**b**: Difference shown on the relative scale.

**ftcompcsg
**

Figure 5 Synthetic shot gather in the *f-k* domain.

**a**: Original dataset.

**b**: Compressed/decompressed dataset.

**ftcompcsgdiff
**

Figure 6 Difference of the two results in last figure. High frequency noise is the main component of this figure.

**a**: Difference shown on the absolute scale.

**b**: Difference shown on the relative scale.

After one compress/decompress cycle, random noise is removed from
the input dataset. It means that this technique attains high compression
ratios by filtering uncoherent component.
We conduct a series of tests using different *scale*
values. As shown in Figure 7 and Figure 8, with the
increase of *scale* value, the compression ratio increases more and more
sharply and SNR decreases more and more slowly.

**csgcr
**

Figure 7 Compression ratio increases from 6 to 76, when *scale* value increases from 0.05 to 0.9.

**csgsnr
**

Figure 8 SNR decreases from 32.88dB to 7.11dB, when *scale* value increases from 0.05 to 0.9.

** Next:** Synthetic 3-D data
** Up:** APPLICATION OF WAVELET-BASED COMPRESSION
** Previous:** APPLICATION OF WAVELET-BASED COMPRESSION
Stanford Exploration Project

11/12/1997