基于小波理论的多分辨率多传感器数据融合

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数据融合，本文研究了不波变换的特征，提出基于小波包变换的多分辨率多传感器的数据融合算法，算法不需要把小波系数当成白噪声处理，并一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，同时采用双正交小波包变换，这可以克服基于正交小波包变换的多尺度滤波中正交小波因不具有线性相而产生恢复失真的缺陷，进一步提高滤波性能。

Multiresolution Data Fusing Based on Wavelet Theory

Wavelet transform is very suitable for processing multiscale signals, and it can be used in multiresolution data fusion. In this paper algorithms based on wavelet packets for multiresolution data fusion are proposed, which are efficient and highly parallelizable. So it avoids big matrix computations, and also has effective performances without assuming wavelet coefficients to be white noises. Because of the character of linear phase of biorthogonal wavelet, the algorithm based on biorthogonal wavelet packets can overcome the drawback of signal distortion caused by orthogonal wavelet packets transform which is nonlinear phase. Simulation verifies that the performance of multiscale data filtering is improved through the utility of biorthogonal wavelet packets transform.