Study On Signal Denosing Based On The Adaptive Algorithm

Signal processing has many aspects, including coding, transmission, detection and recognition, etc. In this process, inevitably accompanied by a certain noise, noise jamming signal communication, destroy the essential characteristics of signal, therefore, to remove or weaken the noise signal in the useful signal detection and extraction is very important. The traditional theory of signal denoising was based on the gaussian noise background, at the moment, the equipment complexity is very high already can no longer be simply considered as gaussian noise. The nature of the non gauss noise and gauss noise, the research methods and engineering application are different. In this paper mainly studies a class of non-gaussian noise, dual mode noise denoising, is the development and perfection of the classical theory.Firstly, briefly describes the difference between the dual mode noise and gaussian noise, and then established the mathematical model of three types of dual mode noise, analyze the probability density function, and points out its statistical properties. After that, detailed introduces the basic knowledge of wavelet analysis and neural network, to clarify the content of their respective characteristics, application and improvement, etc. Lay a solid theoretical foundation for wavelet threshold and wavelet neural network denoising.Secondly, analysis of the wavelet threshold denoising principle, algorithm and steps, points out its advantages and disadvantages, then put forward adaptive wavelet algorithm and expounds its superiority systematically. With classic algorithm and adaptive wavelet algorithm contrast experiment was carried out, the simulation shows that the adaptive wavelet algorithm is compared with the classical algorithm, image clear and smooth, the signal-to-noise ratio and the mean square error(mse) have been improved obviously.Finally, on the basis of the idea of hybrid algorithm, combined with the variable gradient vector momentum item and self-adaptive learn rate, adaptive learning algorithm is designed and wavelet neural network is constructed, using adaptive learning algorithm and wavelet packet transform to test denoising of dual mode noise,comparing test image results show that adaptive learning algorithm to dual mode of eliminating noise(weaken) effect is good. The research of this paper is the beneficial attempt of wavelet neural network in the field of signal treatment.