【摘要】 基于DONOHO的软阈值去噪方法,提出了一种改进的小波软阈值算法。该算法利用高斯白噪声的正态分布特点、3Σ法则和小波变换系数的分布特点,在有效去除白噪声的同时较好地保留了信号的局部特征,且不受采样点数的影响。MATLAB仿真表明,该法在较强白噪声干扰下仍能有效去噪,较之无偏风险、固定阈值、启发式阈值、极大极小值等去噪方法更为有效,特别适于强噪声背景下弱信号的提取。更多还原

【Abstract】 An improved wavelet soft-threshold denoising algorithm is proposed on thebasis of the Donoho’s soft-threshold denoising method,whichis composed of the normal distribution characteristic of Gaussian white noise,3σ law and the distributing characteristic of wavelet transform coefficients.The algorithm well reserves the local characteristic of signal while effectively reducing the white noise,and it is not effected by sampling points.The Matlab simulation demonstrates that it is also able to denoise more effectively than other denoising methods(such as the Rigesure method,Squwolog method,Heursure method and maximum method) when the signal is seriously disturbed by Gaussian white noise.Therefore,it is especially suitable for extracting weak signal from seriously disturbed signal.更多还原