Title of Paper: Optimized Compression for Earth Science Data
Using JPEG 2000
Principal Author: Dr. Janet Rountree
Abstract: The space-based and ground-based sensors planned
by the Earth Science Enterprise (ESE) will acquire huge volumes of data that
must be transmitted, stored, and processed. Some reduction in these data sets
is necessary if the end user is to receive a manageable quantity of useful
information. Therefore it is evident that data compression is a key technology
for ESE missions in the near-, mid-, and far-terms.
JPEG 2000, a wavelet-based algorithm, is being promulgated by the
International Standards Organization (ISO) as the next industrial standard for
image data compression. It is both more efficient and more flexible than its
predecessor, JPEG. It performs both lossless and lossy compression at a
user-selectable compression ratio.
Under a previous contract with NASA’s Explorer Technology program, SAIC
developed a "scan-based" (low memory) implementation of JPEG 2000
Part I, the most basic form of the algorithm. JPEG 2000 Part II, which is
currently being finalized by ISO, contains many features of particular interest
for Earth Science applications. These include special wavelet filters and
decomposition trees for SAR data; single-sample overlap wavelets for artifact
reduction; trellis-coded quantization for highest visual quality; and multiple
component decorrelation for hyperspectral data. In this project, we will
incorporate these Part II features into the scan-based implementation of JPEG
2000.
The project will continue by testing the optimized software on Earth
Science data in a laboratory environment. The software will then be ported to a
flight simulation environment and tested there. The goal is to advance the
technology from Technology Readiness Level (TRL) 3 to TRL 5 in three years.