Efficient and Fast Pattern Matching in Stream Time Series Image Data

In the recent years stream time series data management has become an important research area due to its wide variety of real life applications. In this paper, we propose a technique for similarity matching between staiic/dynamic patterns and stream time-series image data. In order to perform effective retrieval we propose a multi scale segment median approximation representation for stream time-series image data, which can be computed with respect to change in temporal events and hence is suitable for cooperating with temporal behavior. In addition, we propose an efficient pruning algorithm over the multiscale representation. Mainly, the pruning scheme is to reduce the search space and to retrieve the similar patterns effectively. Experiments carried out in this work show that the techniques for stream time series image data representation and the level by level pruning scheme can efficiently generate all candidate patterns and perform similarity matching over stream time series image datasets and performs well compared to the existing methods such as multiscale wavelet.