| Summary: | 碩士 === 國立海洋大學 === 航運技術研究所 === 84 === Impusle noise interferes with signals in an unpredictable
occurrence. The performance of an adaptive signal processor is
limited in the environment of impulse noise. The robust
normalized least-mean-square (RNLMS) approach utilizes the
techniques of nonlinear element and LMS adaptive processor to
remove the interference. Due to its poor performance in low
frequency band, it gets worse in the case of wide band signals.
In this thesis, we employ the wavelet transform to subdivide the
signals into high and low frequency bands. The RNLMS and the
median filter are applied to the high and low bands,
respectively. Then we reconstruct the filtered signal. As a
result, it shows the proposed approach outperforms the RNLMS by
computer simulation.
|
|---|
