Research On The Algorithms Of Superresolution Image Reconstruction With Wavelet And Interpolation

The reconstruction algorithms both from single low resolution (LR) image and a sequence of LR images are deeply researchcd in this thesis. In single LR image reconstruction, not only the edge-adaptive interpolation is simulated, but also a new algorithm combining with wavelet transformation and edge-adaptive interpolation for improving the resolution of a given image is presented. The results indicate that the method is superior both in visual effect and in the value of Cross-correlation and Power Signal-to-Noise Ratio. In sequence LR images reconstruction, a combination of pre-filter gradient motion estimation and picking median of given images is proposed. A better reconstruction image was given with less computation. And the noise of images can also be retained effectively. Then a classical algorithm in wavelet domain is stimulated and an improved method with dividing idea is presented. By this method operation quantity is reduced and speed is picked up.