Research Of Noise Detection And Noise Reduction Algorithm For Meteorological Information

Meteorological information,which is closely related with people's daily life,is interfered by certain types of noise inevitably in procedures of emerging and transmission.However,the existing noise detection algorithms are restricted to single type of noise,and certain denoising algorithm is effective to special noise,both of which couldn't fulfill the requirements of noise handling for meteorological images.In view of these problems,this thesis has done research on noise detection algorithms,denoising algorithms and algorithm optimization at three types,called as speckle noise,salt & pepper noise and Gaussian noise.The major work in this paper is as following.Firstly,principles of noise arising,noise detection and denoising algorithm at three types of noise named speckle noise,salt & pepper noise and Gaussian noise,were analyzed,which established a foundation for developing novel algorithms of noise detection as well as denoising.Secondly,aiming at limitations of the existing detection algorithms,a novel multi type's noise detection algorithm is proposed based on Zernike moment,unnoticeable component of high frequency,peak signal to noise ratio(PSNR)and spectral cluster.The new algorithm was consisted of several models like image library,image extracting,characteristic value computation and clustering analysis.Results show that,this algorithm not only holds a higher successful detection rate,but also realizes blind detection of three types of noise.Finally,to solve the problem of that,the certain denoising algorithm was great useful to special type of noise,an associated algorithm was presented based on median filtering and total variation(TV)model.On basis of it,an optimization algorithm of TV model was realized based on duality rule and gradient algorithm.Results show that,the raised algorithm mitigates the degrading of image by noise.Moreover,convergence rate is improved as well.All of the above research on noise detection and denoising not only provide reference for reducing influence of noise to meteorological images,but also has meanings to improve accuracy,consistency and economy in image transmission.