基于小波系数变换的小波阈值去噪算法改进

An improvement of wavelet shrinkage denoising via wavelet coefficient transformation

小波阈值去噪是近年兴起的一种较好的去噪算法,其一关键点在于准确的选取阈值将细节信号和噪声信号区分开来。该文提出了一种算法对小波系数进行变换,将难以区分信号和噪声的区域放大,以利于阈值的选取,从而达到改进小波阈值去噪的目的。通过使用传统阈值去噪算法和该改进算法进行仿真,结果表明改进算法对去噪指标SNR、SME(平均方差)都有所改善。另外该实验也表明改进算法可以更好的去除噪声,且较好的重现原信号的细节特征。

It is crucial to obtain an accurate threshold to distinguish an original signal from a noisy signal in wavelet shrinkage de-noising,an important de-noising algorithm.A new improvement method was proposed.It was to transform wavelet coefficients to enlarge the area where the original signal and the noise were close to each other.The simulation results illustrated that it brings improvement to the indexes of SNR,SME and visual effects.The experiments also demonstrated that it can improve the de-noising efficiency and reserve the details of the original signal.