Performance Analysis of Adaptive Image Denoising Techniques for Different Levels of Wavelet Decomposition using Orthogonal and Compactly Supported Wavelet Families

Ram Paul, Singara Singh Kasana, Rajesh Kumar Gupta

This paper presents performance analysis of image denoising techniques using different orthogonal and compactly supported wavelets functions of various vanishing moments. The wavelet-based methods such as universal thresholding, level-adaptive and subband-adaptive thresholding are compared with the state-of-the-art Wiener filtering. The wavelet coefficients are modeled by the generalized Gaussian distribution random variables within the subbands. A minimal threshold is calculated from the noise standard deviation of the diagonal subband of the first decomposition level. Then the soft thresholding scheme is applied. The procedure of noise reduction is applied with Daubechies, Symlets and Coiflets wavelet functions of different vanishing moment upto forth decomposition levels. Then the efficiency and performance of these image denoising techniques are compared based on their Peak Signal to Noise Ratios and visual perception. The wavelet domain thresholding is evaluated and examines some improvements for different image complexities contaminated by Gaussian noise of various densities.