摘要
针对时序数据相似性搜索面临的高维性问题,提出一种利用按沃尔什序数排列的离散沃尔什变换((DWHT)W)对时序数据进行维归约的方法。(DWHT)W是正交变换,变换矩阵简单,可以应用快速算法,对时序数据有更好的特征提取能力,用其索引时间序列数据在理论上具备非漏报性质。与基于离散傅里叶变换和基于离散沃尔什变换的对比实验表明,该方法可以获得更高的查询效率。
In order to solve the high-dimensional problem in similarity search over time series data, a new method of indexing and similarity searching in time series databases based on （DWHT）a is proposed in this paper. （DWHT）w is orthogonal and can use fast algorithm, and its transformation matrix is very simple. （DWHT）whas cxcellent performance in feature extraction and is used as an efficient dimensionality reduction technique to permit similarity search over large time series databases without false dismissals. Experimental resuhs demonstrate that performance of the proposed method is more efficient than that of DFT（Discrete Fourier Transform） and DWT（Discrete Wavelet Transform）.
出处
《计算机工程》
CAS
CSCD
北大核心
2011年第8期55-57,60,共4页
Computer Engineering
基金
广东省自然科学基金资助项目(06021484
9151009001000007)
广东省科技计划基金资助项目(2008A060201011)
关键词
时间序列
离散沃尔什变换
按沃尔什序数排列
范围查询
近邻查询
time series
Discrete Walsh Transform（DWT）
arranged according to Walsh sequence
range query,
nearest neighbor emery
