摘要

The adaptive wavelet transform has better adaptability to signals local features than normal ones. The definition of interval autocorrelation factors is presented. The local correlation of samples and the local features of original signals can be characterized by them. Under each scale, with the help of the original signal local features, the optimal predictors and updaters of different samples are selected adoptively. Simulation results show that the method can give attention to both the signals smooth parts and the singular parts, and get a better transform result. The method especially adapts to the online preprocessing of signals with more changes of local features.

  • 出版日期2007