Algorithm of super-resolution in images and video sequences applying wavelets based on atomic functions

Abstract

A novel algorithm for super resolution (SR) in images and video sequences is introduced. In difference with conventional approaches, new framework employs the spatial spectral properties of an image or sequence applying wavelet technique. Novel wavelets based on atomic functions used in SR have shown good properties of resolution by extension of an image size up to 4 times in comparison with the original low resolution image. Numerous statistical simulations have demonstrated the effectiveness of the novel technique exposing good subjective visual quality as well as significant improvement of SR image.. 978-1-4244-4480-9/09/$25.00 ©2009 IEEE

Super-Resplution Procedures in Images and Video Sequences Based on Wavelet Atomic Functions

“…The precision of the registration stage is an important for the reconstructed image quality, because sometimes it is better to interpolate a LR image using classical algorithms than to reconstruct a HR image/frame from a set of images applying incorrect motion parameters. In the chapter, the proposed methods are also investigated under criterion of real time implementation, where additionally to restoration quality, the time values needed to reconstruct is considered, so, only fast method in motion estimation and SR are employed here, like "block matching" that commonly is used (Gomeztagle et al, 2009). To compare the robustness of the analyzed methods different test images and video sequences are studied, These image data present various physical characteristics, such as fine details, edges, texture, contrasts, smooth and rough background, etc.…”