摘要
小波变换的多尺度特点非常适合多尺度信号的处理,可以用于多分辨率多传感器滤波.通过研究快速提升法小波变换的特性,提出一种可将估计误差方差最小化的动态分辨率分布式滤波算法,算法不需要把小波系数当成白噪声处理,并且能够有效地降低向量和矩阵维数,减少运算,有较好的滤波性能.同时在不同的分辨率级中,利用快速提升法小波变换作为一种连接信号的桥梁.这种算法也可用于动态多分辨率多传感器数据融合.
Wavelet transform is very suitable for processing multi-scale signals, and it can be used in multiresolution filtering. In this paper, an algorithm for dynamic distributed filtering of multi-resolution multi-sensor is proposed, based on quick-lifting wavelet transform. By avoiding big matrix computation and eliminating the assumption of white noises for wavelet coefficients, it is efficient and highly parallelizable. Simulation results also show that this algorithm improves the performance of multi-scale data filtering. Moreover the quick-lifting wavelet transform is used as a signal-linking bridge between multi-resolution levels. The algorithm can thus be used for dynamic multi-resolution multi-sensor data fusion.
出处
《控制理论与应用》
EI
CAS
CSCD
北大核心
2006年第2期275-278,共4页
Control Theory & Applications
关键词
小波变换
快速提升法
多分辨率
滤波
wavelet transform
quick-lifting
multiresolution
filtering
