An Improved Image Denoising Algorithm

Wavelet theory has been developed rapidly in recent years, which has increasingly wide application in image denoising. It is very important to select threshold function and threshold in wavelet threshold denoising algorithm. Different selections will affect the denoising effect directly. In the paper, traditional soft and hard threshold functions were further analyzed and studied, advantages of denoising performances in both soft and hard threshold functions were combined for proposing an improved threshold function. Threshold proposed by Donoho was improved according to characteristics of wavelet decomposition layers and noise wavelet coefficient in the paper. Different wavelet basis and decomposition layers were compared in order to optimize simulation results in the paper. Then, appropriate wavelet basis and decomposition layers were selected for experiment. The experimental results showed that denoising method proposed in the paper can effectively eliminate the noise in the image with good edge information protection and visual effect.