Super-resolution Image Reconstruction Based On Wavelet Transform

In recent years, with the advancement of science and technology, the growing popularity ofthe network and appearance of multimedia technology, Image reconstruction technologies havebeen applied widely. The technology is a challenging task in the field of modern signal andinformation processing. As we all know, when we want to get an image, imaging modalities andimaging conditions are restricted by physical aspects, imaging system can’t get all theinformation about the original scene. And the image resolution is low, can’t meet the needs ofpeople. So what problem the people commit to solve is how to improve the identification of animage. Because of this background, Super-resolution image reconstruction technology isdeveloping rapidly. The basic method of Super-resolution reconstruction technology is using amethod of signal processing and using this existing low resolution image to get someinformation outside of cut-off frequency without changing those hardware devices. Thetechnology has been frequently used in the field of military, medical and so on. The thesis mainlyworks and achievements are as shown below:First of all, this paper introduces the history of the technology of super resolutionreconstruction, and some knowledge of wavelet transform. The wavelet transform has manyadvantages: it can be multi-scale decomposed and can separate high frequency and lowfrequency information of image. Based on these advantages, the wavelet transform has importantapplications in the field of image analysis.Secondly, this paper studies the single frame super-resolution image reconstruction.Meanwhile, for the defects of the classical algorithms, we propose the optimization algorithmand we applying the wavelet transform, image fusion and image extrapolation technique isintroduced into the algorithm.Finally, we study the multiple image super-resolution reconstruction technique. On thosehigh resolution grids, we use the method of iteration to change the resolution of the image. Dueto the multiple images contains more information than single frame image, so the performancesof multiple images reconstruction technique are greater. Then, after processing of an image, weuse the wavelet soft threshold denoising method to reduce noise interference.The results tell us that our algorithm can overcome some disadvantages of the classicalinterpolation methods and effectively improved the resolution and PSNR of an image.