基于复合评价值的小波去噪参数选定方法

Wavelet Denoising Parameter Selection Method Based on Composite Evaluation Value

针对小波阈值去噪法应用到含噪信号时小波基和分层数难以统一确定的问题,提出了一种融合传统评价指标对于小波去噪最优参数选定的方法。根据一般评价指标的意义和小波分解产生的细节信息及逼近信息的意义,融合得到复合评价值,通过复合指标值寻找小波去噪后的极值点来确定最优小波基函数和分层数。通过对blocks和bumps信号的实验结果与其他文献的对比,证明了该方法的合理性和有效性。最后,将该方法应用到超声回波信号去噪中,在选取sym7小波基和4层分层数时,得到的复合评价值最小,超声回波信号取得最优去噪效果。

Aiming at the problem that the wavelet generating function and decomposition levels are difficult to be determined uniformly when the wavelet threshold denoising method is used,a fusion evaluation value is proposed for the optimal parameter selection.According to the meaning of the general evaluation index and the details of the wavelet decomposition and the meaning of the approximation information,the composite evaluation value is obtained by fusion.The optimal wavelet basis function and the number of layers are determined by finding the extreme points of the wavelet denoising by the composite index value.The result of experimental blocks and bumps signal,compared with other reference,proves the rationality and superiority of the method presented in this paper.Finally,the method is applied to the ultrasonic echo signal denoising.When the sym7 wavelet base and the 4-layer hierarchical number are selected,the composite evaluation value is the smallest,and the ultrasonic echo signal obtains the optimal denoising effect.