Image Restoration Research On Frequency Domain And Wavelet Domain

Image restoration is one of the most important issues in digital image processing, it has been widely used in various fields of science and engineering. In the process of image restoration it is required to remove the noise while preserving the details, which seems to be a contradiction. Because of the edge blurry brought by noise removement, the frequency-domain image restoration could not offer us a satisfying result. The wavelet method is with high flexibility in time-domain, which is able to represent the edges and catastrophe points. The wavelet method can offer a better result in keeping image edge details. At present, the wavelet image restoration is a hot research topic.An in-depth research of various image restoration methods has been carried out, and a restoration method using frequency-domain normalized iterative algorithm combined with wavelet-domain Wiener filting is introduced based on the features of those methods. The shortcoming of the wavelet method that having a better performance while complex in computing is made up, as well as the shortcoming of the frequency method that simple in computing while lower in performance.. The algorithm analysis and simulation result indicate that this new algorithm has brought a impressive improvement in MSE and ISNR criteria compared with mere frequency-domain method and wavelet method, and has a great theory and practice significance to image restoration algorithm.