Similarity search over time series data using DCT
基于DCT的时序数据相似性搜索

High dimensionality is the main difficulty of similarity search over time-series data. The most promising solution involves performing dimensionality reduction on the data, then indexing the reduced data with a spatial method. Recently, two methods of dimensionality reductions have been proposed, DIrT and DWT. In this paper we proposed a new method, dimensionality reduction with DCT, and further provided the method of similarity search about range query and nearest neighbor query. Compared with those methods based on DFT and DWT, it is more efficient in theory and experiment.