Image super-resolution by prediction of dual tree-CWT coefficient at a finer scale

This paper presents an image of Super-Resolution (SR) technique by the construction of DT-CWT coefficients for a larger scale from the information at a smaller scale for all the subbands. The DT-CWT coefficients prediction for each subband of an image at a finer level is based on phase prediction and estimation of the magnitude separately, followed by combining the magnitude and phase. Inverse DT-CWT is taken with the coefficients at a finer level of each subband along with a Low-Resolution (LR) image in place of a low subband to reconstruct a high-resolution image. The proposed technique is applied to various images, including satellite and standard images. The quantitative and visual results have established the superiority of the proposed scheme over conventional and various state-of-the-art techniques.