Title :
Study of fast super-resolution reconstruction method
Author_Institution :
Dept. of Comput. Sci. & Eng., East China Univ. of Political Sci. & Law, Shanghai, China
Abstract :
An improvement POCS method to enhance the image resolution is proposed in this paper. Traditional POCS method obtained the reconstruction by iterating pixel-to-pixel with high computation. The algorithm treats each LR image as a whole set and uses subpixel registration to position each LR pixel that mapped in the HR grids. We define the residual as the norm of the differences between the consecutive estimated HR image pixels and iterate the reconstruction until the residual is below an appointed value. A number of experiments using simulated and real multiple images demonstrate the efficacy and computational superiority of the proposed algorithm.
Keywords :
image reconstruction; image registration; image resolution; HR image pixel; LR image; POCS method; image resolution; project on convex set method; subpixel registration; super resolution reconstruction method; Image reconstruction; Interpolation; Pixel; Signal processing algorithms; Signal resolution; Spatial resolution; POCS; mutiple images; super-resolution;
Conference_Titel :
Computer Science and Automation Engineering (CSAE), 2011 IEEE International Conference on
Conference_Location :
Shanghai
Print_ISBN :
978-1-4244-8727-1
DOI :
10.1109/CSAE.2011.5952824