A novel threshold method for 3-band wavelet de-noising

This paper proposes a novel threshold method based on the penalized threshold presented by Birge and Massart, the novel threshold is derived in 3-band wavelet domain. Instead of selecting the coefficients which are larger than some threshold, the method merely fixes the number of coefficients we want to keep equal to some value and takes the largest ones. It is effective because the number of wavelet coefficients we kept at each level can be calculated directly and it is adaptive as these values depend on some variables such as alpha and C₁, which can be various for particular signal to get the best performance. The experimental results show that in the 3-band wavelet domain, the proposed method represents better characteristics than D.J method in protecting details and improving the signal-to-noise ratio of reconstructed signal. Moreover, the application of the proposed method to simulated ultrasonic signal is demonstrated in the experiment.