SIMILAR LARGE OBJECTS IN DATABASES
 | Maude Manouvrier Suprevisor: Mme Geneviève Jomier Equipe Bases de Données et Génie Logiciel Laboratoire LAMSADE The defense of the thesis was done in January 2000! You can download the files of my thesis (written in french). |
This thesis deals with the problems of large object versions in databases. Images are large objects loosely structured. Quad-tree may encode them. In image processing, users need to store and manage efficiently images resulting of processing operations. All along the process, users may consider the created images as versions of an original image. In this thesis, a structure, called Generic Quad-Tree, store similar images. Two images are defined similar if, for both images, the quad-trees that encode them are similar, i.e. differ only for a relative small number of nodes. The Generic Quad-Tree optimizes the memory space of similar images and allows processing of operations on them, like images comparison, comparison of the same area in different images, or simultaneous updates in different images.
Large objects highly structured obviously appear in OODBMS, for instance during the management of collections of objects or roots (set of object identifiers). Index structures, for instance B+ tree, often manage this kind of objects. In order to solve the redundancy problem of set versions or collection versions, we propose mechanisms to manage B+ tree versions. The proposed mechanisms optimize the management of this kind of objects and may be generalized to any index tree structure. Therefore, they apply to a large domain of applications and particularly Geographic Information Systems and datawarehouse domains.
Keywords :
database, index, large object loosely or highly structured, versions, storage optimization, image similarity, collection of objects
My CV
