抑制震荡的自适应小波阈值去噪方法

Ringing Artifacts Reduction in Adaptive Wavelet Threshold Denoising

分析了小波阈值去噪方法在信号的间断附近产生震荡的原因,根据产生机理的不同将它们分为属于伪Gibbs现象的震荡和由于小波阈值去噪在信号的间断部分因为去噪能力下降而属于噪声残留的震荡。为了消除这些震荡,结合软、硬阈值方法的特点提出了一种新的阈值函数自适应地处理小波系数。试验表明,该方法有效地抑制了信号间断点处的震荡,获得了优于传统阈值方法的去噪效果。

The reason of ringing artifacts appearing near the discontinue points in denoised signal is analyzed by traditional wavelet shrinkage methods. Based on the different origins, the ringing artifacts are devided into two classes One class is Pseudo-Gibbs phenomena and the other class is the remnant noise. To reduce these artifacts, a new threshold function and a new adaptive algorithm are obtained which could handle wavelet coefficients flexibly. Numerical experiments show that presented method can significantly depress ringing artifacts.