Decision of threshold for singular value decomposition filter based on SNR''s empirical value
基于信噪比经验值的奇异值分解滤波门限确定*

Aiming at shortcomings of the existing singular value truncation criterion of singular value decomposition filter (SVDF), this paper put forward a new method to decide the threshold for SVDF, based on the empirical value of signal to noise ratio(SNR). It deduced the mathematical relation between the threshold and SNR. Finally, approved the rationality and validity of the promoted noise-elimination algorithm with SVDF by the experiment results. Compared to the existing methods, the effect of noise elimination is obviously improved, with clear principle and easy implement.