Ece497kj Course Project Applications of Wiener Filtering in Image and Video De-noising

Image and video signals are sometimes corrupted by Gaussian noise. Wiener ltering could reduce noise eeectively. Wiener lter is the best linear restoration lter in the sense of mean square error (MSE). In this project, Wiener lters were implemented to reduce noise in noisy images and video sequences. The lter taps of Wiener lter adapt with local signal and noise statistics. This report has four sections. In section I, concepts of Wiener lter were discussed. If there is no blurring eeect, and additive noise is Gaussian, the simple form of 1-tap Wiener lter can be justiied. In section II, Wiener lter was implemented to process noisy still images. Simulations were done in both spacial domain and wavelet domain. In section III, Wiener ltering was extended to reduce noise in video sequences. Since video sequence is 3-D, there were several conngurations that Wiener ltering could be implemented. Connguartions were disussed, and simulations results were presented. Further work was proposed and discussed in section IV.