Histogram distance for similarity search in large time series database

Abstract

Dynamic Time Warping (DTW) has been widely used for measuring the distance between the two time series, but its computational complexity is too high to be directly applied to similarity search in large databases. In this paper, we propose a new approach to deal with this problem. It builds the filtering process based on histogram distance, using mean value to mark the trend of points in every segment and counting different binary bits to select the candidate sequences. Therefore, it produces a more appropriate collection of candidates than original binary histograms in less time, guaranteeing no false dismissals. The results of simulation experiments prove us that the new method exceeds the original one. © 2010 Springer-Verlag Berlin Heidelberg.