在运用小波阈值法去噪时,在一些不连续点处有时会产生伪吉布斯现象。为了去除这种现象,考虑在阈值法的基础上对信号进行循环平移,然后再对平移后阈值上的系数进行放大,对信号进行消噪处理。通过仿真实验,可知该方法消噪后的视觉特性较好,并且信噪比提高,均方根误差有所降低,从而说明了该方法的有效性。
In some case, such as on the discontinuities points, the Gibbs phenomenon will exhibit when we use wavelet thresholding to remove noise of signals. In order to remove this phenomenon, we consider the "cycle spinning" method based on threshold wavelet and then use the coefficient magnify to remove the signal noise. The results of the experiment show that the vision of deniosing is better and the RMSE of signal has been decreased a lot while the SNR has been increased, which indicates the method presented in this paper is effective.