Spatial Temporal Databases

This research project will first develop efficient indexing strategies for spatio-temporal data sets. This will include access methods for future spatio-temporal queries as well as access methods for historical spatio-temporal queries. The next step will be developing mining techniques for spatio-temporal databases. Finally the project will generate spatio-temporal data sets for experimentation.

Our long-term goal is to develop a database system that manages spatio-temporal data efficiently and effectively. We believe that using the current and projected advances in hardware and network technology, we will be able to build a system that can achieve high performance and reduced cost for spatio-temporal database applications. In the context of the proposed research effort, we plan to provide tools and methods that will be used as building blocks for the system that we envision. In particular, we will concentrate on the following problems:
(a) Efficient Indexing of Spatio-temporal Datasets: There are at least two kinds of interesting queries in such an environment, namely "Future" and "Historical" queries. If the functions by which object move/change are known, we can answer queries about the objects' anticipated positions/extent in the future. The answer to such queries is based on the time the query is executed. We will consider indexing schemes for range, neighbour and aggregation future queries. On the other hand, if the past locations of moving objects are stored in a database, we are interested in providing efficient index structures for querying the past.
(b) Mining Spatio-temporal Databases: Spatio-temporal data are usually noisy and complex. We plan to investigate methods and algorithms for efficient computation of appropriate similarity models and data mining operations (like similarity indexing) in this environment.
(c) Generating Spatio-temporal datasets: For benchmarking, evaluating and comparing spatio-temporal systems we need robust methods for generation of realistic workloads. We will identify important properties of real datasets and investigate how we can use them to generate useful datasets for experimentation.

Data Mining in Video Databases

The aim of this research project is to develop and test methods for indexing, retrieval, and data mining of human motion trajectories in video databases. Computer vision techniques are being devised for automatic extraction of human motion time series data from video. Algorithms are being developed that can be used to discover clusters and other patterns in the extracted motion time-series data. One promising direction being explored is to model the observed motion time series sequences with a finite mixture of Hidden Markov Models (HMMs). Use of the HMM representation presents certain advantages with regard to modeling; however, it presents important challenges for the design of efficient clustering, indexing, and retrieval algorithms. Thus more efficient, sampling-based and embedding-based methods must be formulated. The products of this research effort can enable numerous applications that are valuable to society: homeland security; video-based analysis of human biomechanics for occupational safety, as well as dance and sports training; archive management and analysis for news, entertainment, and sports video; and video database management for non-intrusive monitoring of the motion patterns of handicapped, infirm, or elderly people to detect decline, danger, or to alert caregivers when needed.