Digital Image Restoration Research Based On Wavelet-domain

Image restoration is an important research field in digital image processing and its research findings are applied in many kinds of science and engineering fields. It's required that the restored image be get rid of noise and the edge and the image detail be preserved. But they are always contradictive. Wavelet bases have flexible time-frequency dilation capability, which can characterize the image with edges and discontinuity very effectively. Nowadays, digital image restoration research based on wavelet analysis is a hotspot.Introduction reviews the progress of image restoration in recent years and introduces the main job of this paper.Chapter 1 works over the theory of wavelet transform. It stresses multiresolution analysis and fast wavelet transform.Chapter 2 studies and analyzes some classical image restoration algorithms, including the frequency — domain algorithms, linear algebra algorithms and parameter model algorithms. It analyzes the theory of the algorithms in detail and expatiates on their advantages and disadvantages.Chapter 3 studies and analyzes the wavelet-based image restoration algorithms , including threshold deconvolution algorithm, iterative regularization algorithm and parameter model algorithm. These algorithms have better performance than classical restoration algorithms in image edge-preserving.Chapter 4 presents wavelet-domain combining frequency-domain algorithms according to the complementation of wavelet-domain algorithms and frequency-domain algorithms, in which the former are better edge-preserving but have complicated calculation and the latter have simple calculation but are worse edge-preserving. The algorithm analysis and the simulation show that the algorithm is obviously superior to the simple frequency-domain algorithm and the wavelet-domain algorithm in visual effect and the performance of MSE and ISNR.End of the paper summarizes and has prospect to the future research orientation.