Lossy compression of scientific data via wavelets and vector quantization
Goldschneider, Jill R
MetadataShow full item record
The high volume of scientific data to be collected from Earth Observing System instruments will require large transmission bandwidth and storage capabilities for archiving. Lossy compression techniques can be used to compress the data, although by definition the original data cannot be recovered. The compression techniques used must ensure that the compressed data have both high visual fidelity and that calculations using the compressed data yield accurate results. In this dissertation, lossy compression algorithms based on the wavelet transform and vector quantization are developed. Optimal bit-allocation algorithms based on pruned tree-structured vector quantization techniques are developed for the discrete wavelet transform and for the more general discrete wavelet packet transform. These algorithms systematically find all quantizers on the lower convex hull of the rate-distortion curve, while for the wavelet packet transform, simultaneously selecting the best basis. The effects of such compression on USC database images is examined for benchmarking purposes, and the benefits of compressing ocean science acoustic sonar data used for the scientific analyses of target detection and temperature analysis are studied.
- Electrical engineering