A New Gaussian Noise Filtering Algorithm

王小兵,孙久运

DOI: https://doi.org/10.3969/j.issn.1009-4822.2011.05.027

2011-01-01

Abstract:In order to filter the Gaussian noise of digital image more effectively,a new filtering algorithm was proposed.Firstly,the image which contains Gaussian noise was decomposed two-dimensional wavelet,obtaining high frequency and low frequency wavelet coefficient.Secondly,retain the low-frequency wavelet coefficient unchanged,at the same time the high-frequency wavelet coefficient was conducted wiener filtering,the wavelet coefficients were reconstructed.Thirdly,the reconstruction image was implemented multi-scale wavelet decomposition,setting new threshold discriminant function so as to weaken unimportant decomposition coefficients.Finally,wavelet decomposition coefficients were reconstructed.By using this filtering algorithm,wiener filtering,wavelet threshold value method and average filtering on Gaussian noise denoising processing.The experiments show that PSNR value of the filtering algorithm is much higher than the other three methods.