An index structure for moving objects in video databases

Abstract

Modeling moving objects and Iiandling various types of motion queries are interesting topics to investigate in the area of video databases. In one type of motion queries, motion of multiple objects is specified by the changes in relative spatial positions of objects. Answering such kind of queries, that involve motion of multiple objects whose identifications cire not specified, requires some type of indexing because the time complexity of processing such a query in the absence of an index structure is 0{N \l{N — n)!), where N is the number of objects in the database and n is the number of objects in the query. In this work, we propose a spatio-temporal index structure, which we call ,S'M/A7-index, and compare its performance against a similar scheme proposed in [18]. The scheme presented in [18] consists of a constraint satisfaction algorithm, which is called Join Window Reduction (JW R ), combined with a spatial index structure (R*- tree). Experimental results indicate thcit SMIST-'mdex outperforms the JW R algorithm. Also, SMIST-'mdex is shown to be scalable to increasing number of frames and objects.