Storage management for vector/raster data in spatial database systems

Abstract

It has been recognized that relational database management systems do not well handle spatial data efficiently. The underlying access methods, storage managers, and query languages of the conventional systems are designed to deal with only simple data types like integer, real number, and string. However, non-traditional complex applications such as geographic information systems, computer-aided design, and computer vision are basically required to manage vector data including points, lines, and rectangles and large raster data like voice, image, and video. In this thesis, for the efficient manipulation of vector data and very large raster data, access method, storage manager, and query language are addressed. For access method, we propose a new dynamic access method, called Binary divided Region (BR) tree, for uniform vector data and then implement search, insertion, deletion, and splitting algorithms used for the BR tree. The BR tree splits a region into two sub-regions of equal size recursively and prevents overlapping of rectangles in internal nodes by clipping objects which cause overlapping region. The proposed access method is compared with the two representative spatial access methods, R-tree and $R^+$-tree, through performance results from analytical study and simulation approach for a VLSI data. For search, when the size of a user-specified rectangular region, called query window, is relatively small, BR tree produces the best performance with respect to the number of disk accesses since it generates the smallest number of nodes and avoids overlapping of rectangles. For storage manager, we propose a new database storage manager, called Buddy-size Segment Storage System (BSSS), to handle large raster data of any size and then implement search, insertion, append, and deletion algorithms used for the storage structure. The internal nodes of the proposed storage manager are identical to the ones proposed in Exodus. However, unlike Exodus that has fixed-s...