中文说明: SLTF算法的原理是:信号通过SLTF算法,在变换域中以比较少的,能量较高的,集中的系数来表示,而噪声通过SLTF算法,在变换域中以比较均匀的,能量较低的,分散的系数来表示。然后,通过设定一个噪声阀值,高于该阀值的系数保留,低于该阀值的系数去掉,这样可有效地消除噪声部分。 SLTF算法具有良好的降噪性能,缺点是在设阀值消噪过程中,会损失部分信号的细节,导致最后还原的时候,信号有一定程度的失真。 降噪的关键在于找到能够在a矩阵中分离噪声能量系数与信号能量系数的变换域,据试验知,M=N=8和M=N=10是具有很好的分离噪声能量系数与信号能量系数的变换域,通过这两个变换域降噪后,能使信噪比提升3db左右。
English Description:
&The principle of sltf algorithm is: signal is represented by less, higher energy and concentrated coefficients in transform domain by sltf algorithm, while noise is represented by more uniform, lower energy and scattered coefficients in transform domain by sltf algorithm. Then, by setting a noise threshold, the coefficients higher than the threshold are retained, and the coefficients lower than the threshold are removed, which can effectively eliminate the noise part.