Title of Paper: Wavelet Compression of MODIS Satellite Images
Principal Author: Dr. Richard Ladner
Abstract: Moderate
Resolution Imaging Spectroradiometer (MODIS) data are collected from the Terra
satellite at a rate of 11mb/s. These data are to be transmitted to earth,
stored, and analyzed by various algorithms. One solution to reducing the
bandwidth and storage needs for these data is to apply wavelet based
compression. Wavelet-based compression algorithms provide a tradeoff between
compression ratio and the error introduced in the reconstructed data. The
question is which wavelet coders to use and what compression ratios to use to
avoid distorting significantly the analysis algorithms' results. The
wavelet-based compression algorithm "Set Partitioning in Hierarchical
Trees" (SPIHT) seems to be very well suited to an FPGA hardware
implementation and is known to have excellent compression performance. We
present a controlled study that shows the effect of data compression on MODIS
data. Cloud cover calculations can tolerate very large compression ratios. The
correlation between frequency bands can be used to compress even more.