The HRI method is a multi step index for interval sequences which are generated out of spatial objects via space filling curves. Interval sequences representing high resolution spatially extended objects often consist of very short intervals connected by short gaps. The HRI method groups the detailed black intervals together to longer grey intervals and stores them in a BLOB. A disadvantage of this method is the fact that the storage requirements for the grey intervals table grows with the gap length. Because increasing the gap length, at least up to a certain limit, reduces the number of the grey intervals which positively affects the query response time, large gap lengths are used ignoring the sub-optimal storage requirements.
In this master thesis, several algorithms for BLOB compression are presented. First, two versions of a high performance data compressors are discussed which keep the overall size of a CAD test data set almost constant, although the gap length is increased by an order of magnitude. The second approach uses the fact that BLOBs often contain lots of repeating structures, also called patterns. A pattern decomposition method is introduced which stores frequent parts of an object in a separate BLOB to reduce space.
Bearbeiter | Peter Kunath |
Betreuer | Martin Pfeifle |