基于提升静态小波包变换的自适应消噪方法

Investigation of a denoising method based on lifting static wavelet packet transformation

基于离散小波变换(DWT)的自适应消噪方法为雷达信号的滤波提供了一种可行的方法.但DWT不具有平移不变性,若不用相同的小波对滤波后的信号进行重构,则会带来较大的重构误差.针对这一现象,提出了一种基于提升静态小波包变换的自适应消噪方法,它推导了静态小波包的提升实现方法,并设计出适合该系统的确定最优小波包分解树的相应步骤,利用引入更多动量因子的权系数迭代公式对各子带进行自适应匹配,并将匹配结果二次自适应,得到拟合的原信号.仿真结果表明,该方法可在计算量增加不大的前提下,进一步改善系统的滤波性能.

The method of adaptive denoising based on discrete wavelet transform （DWT） provides a feasible solution for radar signal filtering, but DWT does not have the characteristics of translation invariance of the wavelet coefficients. If signal reconstruction is not done with the same wavelet, it will cause considerable reconstruction errors. To deal with this phenomenon, this paper proposes an adaptive denoising method based on lifting static wavelet packet transformation. The authors developed the lifting method for static wavelet packets, and designed corresponding steps to determine optimal wavelet packet trees suitable for this system. They then conducted adaptive matching to each sub-band using a weighted coefficient iterative formula which has more momentum factors. Finally, a second adaptive filter of the matching results was used to acquire the fitted signal. Simulation results show that the method improves filtering performance without substantially increasing calculations.