Comparison of Fourier and Wavelet Transformations for Noise Reduction in Images by Filter Banks

In this paper, some kinds of image noise are introduced and PSNR criteria is used as a tool to compare noise reduction. Beside this, visual comparison is also utilized for Barbara’s image (256*256). Three stages of wavelet decomposition are considered. Symlet-8, Daubechies-8 and Koiflet-2 are used for noise reduction. In Fourier domain, three kinds of filters (averaging, Winer and median) is applied. Symlet-8 shows the best performance but the other three curves show almost the same trend. All three kinds of wavelet filters show better performance than Winer filter for gaussian and speckle noises, but the result is vice versa for salt and pepper noise. In the cases that visual comparison or PSNR criteria are not that much effective for comparison, the difference between original and filtered images is used as the forth image for comparison.

کلیدواژه ها:

image noise reduction ، PSNR criteria ، visual ، comparison ، wavelet transformation