ABSTRACT
This chapter presents a hybrid Fourier-EMD Regularized Deconvolution (For-EMDRD) algorithm. When the impulse response is estimated, the Fourier shrinkage and EMD shrinkage follow the ideal deconvolution result. Simulated and experimental results show that For-EMDRD not only avoidsillposed problem in deconvolution, but also improves signal edge character and Signal-to-Noise Ratio. For-EMDRD can be applied in signal estimation of acoustic wideband pulse measurement, removing the effect of transient response of a loudspeaker, eliminating the noise effect, which makes a better separation of the incident wave and reflected wave and improving the measurement accuracy of absorption coefficient. The most used regularized deconvolution for stationary signal is the Wiener deconvolution. Wiener deconvolution can be deemed as a type of shrinkage of ideal deconvolution in Fourier domain. Wiener deconvolution is closer to Fourier-EMDRD. Regularization for deconvolution can avoid the effect of noise to zero point of impulse response. However, deconvolution results are sensitive to regularization parameter.
