In order to effectively measure the correlation velocity in the background of noise and reverberation, researchers from the Institute of Electronic Material, High-Tech Research and Development Center has proposed a method to enhance the noise robustness of correlation velocity measurement in wavelet domain, and they studied the work to evaluate the performance of the new method for correlation velocity measurement compared with classical methods through the simulation experiment.
The new method of correlation velocity measurement is based on background denoising using discrete wavelet transform (DWT), which consists of denoising step, correlation matrix calculation step and correlation velocity estimation step. The white Gaussian denoising step of the seabed reverberation signal using DWT is as follows.
Step 1: We perform the DWT for the bottom reverberation signal received by the receiver.
Step 2: To remove white Gaussian noise, we perform thresholding on the discrete wavelet coefficients.
In order to reduce a white Gaussian noise in received signal of correlation velocity log (CVL) and to enhance the seabed reverberation signal, we perform the thresholding for discrete wavelet coefficients. In order to minimize the information loss of seabed reverberation signal and improve the wavelet denoising performance, we use a threshold function compromised between Garrote and Hyperbola threshold functions.
Step 3: Denoising signal is reconstructed by inverse wavelet transform to thresholding discrete wavelet coefficients.
Experimental results showed that the new method had a high accuracy of correlation velocity measurement and improved the noise robustness by more than about 6dB compared to the classical methods. We improved the accuracy of correlation velocity measurement and the noise robustness by introducing this method into the CVL.
The research results were published in the journal "International Journal of Wavelet, Multiresolution and Information Processing" under the title of "A method to enhance the noise robustness of correlation velocity measurement using discrete wavelet transform"(https://doi.org/10.1142/S0219691321500491).
