Research On Homomorphic Filtering Techology In Signal Processing Application

In signal processing field, there is one kind of filtering problem. The mixed signal iscomposed of original signal and noise, which is in the form of product or convolution. Thenthe ordinary linear filtering method doesn’t work. In recent years, many scholars haveproposed wavelet transform, adaptive filtering and other algorithm to work this problem out.In this paper, homomorphic filtering technology is applied to achieve this kind of signalfiltering and signal-to-noise separated processing.In the beginning of this paper, the definition of homomorphic system and its structure isdiscussed, then the concept of complex cepstrum is introduced, its properties and calculationmethod is also discussed here. The phase compensation method is used to solve themulti-valued problem of the complex cepstrum.In the middle part of this paper, two types of homomorphism filtering system isintroduced, their structures and applications are discussed. For multiplicative homomorphicsystem, there are two specific applications, namely: track FSK signal mixed withmultiplicative noise interference; the product factors of the image (the illuminationcomponent and the reflection component) are not reasonably configured, which caused theuneven illuminated problem. The homomorphic filtering technology is applied respectively onthe track noise signal and uneven illuminated image, which realized the signal-noiseseparation and contrast correction. From the result of filtering, the separation effect of thesignal-to-noise is good, and the image becomes more clear and illuminated uniformly. In theconvolutional homomorphic system, two kinds of underwater acoustic channel are used assignal transmitted environment, according to the reverberation problem of the signal, which istransmitted in the multipath underwater acoustic channel, the homomorphic filteringtechnology is used to extract the direct path wave from the reverberation wave. The error ofthe restored signal is small, but this method needs higher signal-to-noise ratio.In the end of this paper, two specific examples of engineering application are introduced,and the homomorphic system is put forward to improve estimation performance. The firstapplication, the generalized cepstrum correlation method is used to realize the effect ofextracting time delay, the time delay estimation accuracy of this method is higher than thetraditional cross-correlation method, which provides a guarantee to the underwater accuratelypositioning; The second application, the cepstrum method is used to target recognition, whichmethod extracted the characteristic information from the ship radiation noise, then itsanti-noise performance is proved to be better than the commonly used power spectrummethod.