Search

This paper suggested a de-noising algorithm used in grayscale images. As long as the noisy image does not give the desired view of its features, de-noising is required. The algorithm is based on block matching and wavelet transformation. Euclidean distance for blocks similarity is exploited, which demonstrate more accurate in finding similar blocks depending on soft thresholding. Regarding wavelet transform, a combine of hard thresholding is performed for HH and LH sub-bands while soft thresholding is used in LL and HL sub-bands of the decomposed images. Three types of noise is encountered: Gaussian noise, salt & pepper noise and speckle noise. The measurements are employed to evaluate our work is MSE and PSNR and SSIM. Finally a comparison of the results shows that our method outperforms traditional wavelet using hard or soft thresholding.

This paper suggested a de-noising algorithm used in grayscale images. As long as the noisy image does not give the desired view of its features, de-noising is required. The algorithm is based on block matching and wavelet transformation. Euclidean distance for blocks similarity is exploited, which demonstrate more accurate in finding similar blocks depending on soft thresholding. Regarding wavelet transform, a combine of hard thresholding is performed for HH and LH sub-bands while soft thresholding is used in LL and HL sub-bands of the decomposed images. Three types of noise is encountered: Gaussian noise, salt & pepper noise and speckle noise. The measurements are employed to evaluate our work is MSE and PSNR and SSIM. Finally a comparison of the results shows that our method outperforms traditional wavelet using hard or soft thresholding http://dx.doi.org/10.25130/tjps.23.2018.038