基于提升小波的超声信号降噪方法

Ultrasonic Signal Denoising Based on Lifting Wavelet

针对经典小波在逼近信号时产生的细节信号会被当做噪声滤除的问题,提出了基于提升小波的超声信号降噪方法。该方法用剖分、预测、更新,把信号进行单层分解,然后把分解后的高频系数和低频系数同时进行阈值函数处理,再进行小波重构。相对于传统的一代小波,该算法不依赖于傅里叶变换,计算简单,可有效节约内存。仿真与实验结果表明:该算法可以有效地提高信噪比,降低原始信号中所含有的噪声,并可以起到数据压缩的效果。

When classical wavelet reconstructs signals,the details of signals will be filtered as noises.To solve the problem,a method based on lifting wavelet theory for ultrasonic signal denoising was proposed in this paper.The signals single-layered decomposition was carried out through splitting,predicting and updating.The obtained high-frequency and low-frequency coefficients were transformed with threshold function after decomposition,then,the signals were reconstructed.Finally,the noise was effectively eliminated from original signals.Compared with traditional generation wavelets,this algorithm was suitable for real-time DSP processing due to its brevity and memory efficiency.Simulation results showed that the algorithm could effectively improve the signal-to-noise ratio,reduce the noise level in original signals,and had an obvious effect on data compression.