论文部分内容阅读
提出了分形信号的小波分解与重构的一种快速算法。针对分形信号的自相似和长时相关的特点，采用离散小波变换（DWT）对分形信号进行多尺度分解，使其成为各尺度上的近似平稳信号，从而可利用通常的Wiener滤波或Kalman滤波方法进行估计，然后再由DWT进行多尺度重构，估计出被噪声污染了的原始信号。重点对DWT的滤波过程进行算法设计，并估计了计算复杂度。
A fast algorithm of wavelet decomposition and reconstruction of fractal signals is proposed. Aiming at the characteristics of self-similarity and long-term correlation of fractal signals, the discrete wavelet transform (DWT) is used to decompose the fractal signals into multi-scale and make them approximate smooth signals at each scale. Thus, the usual Wiener or Kalman filter Estimate, and then multi-scale reconstructed by DWT to estimate the original signal contaminated by noise. Focus on DWT filtering algorithm design, and estimated the computational complexity.