Date of Award


Publication Type

Master Thesis

Degree Name



Electrical and Computer Engineering

First Advisor

Jonathan Wu


Applied sciences




The estimation of some unknown quantity information from known observable information can be viewed as a specific statistical process which needs an extra source of information prediction strategy. In this regard, image super-resolution is an important application In this thesis, we proposed a new image interpolation method based on Redundant Discrete Wavelet Transform (RDWT) and self-adaptive processes in which edge direction details are considered to solve single-frame image super-resolution task. Information about sharp variations, both in horizontal and vertical directions derived from wavelet transform sub-bands are considered, followed by detection and modification of the aliasing part in the preliminary output in order to increase the visual effect. By exploiting fundamental properties of images such as property of edge direction, different parts of the source image are considered separately in order to predict the vertical and horizontal details accurately, helping to consummate the whole framework in reconstructing the high-resolution image. Extensive tests of the proposed method show that both objective quality (PSNR) and subjective quality are obviously improved compared to several other state-of-the-art methods. And this work also leaved capacious space for further research, not only theoretical but also practical. Some of the related research applications based on this algorithm strategy are also briefly introduced.