Title :
A new method of images super-resolution restoration by neural networks
Author :
Zhang, Liming ; Pan, Fengzhi
Author_Institution :
Dept. of Electron. Eng., Fudan Univ., Shanghai, China
Abstract :
Super-resolution restoration from the tow-resolution image is an ill-posed problem if there´s no assumption. This paper proposes a new super-resolution scheme based on combining neural network with classical interpolation algorithm. It is shown that our method has better performance than existing interpolation algorithms on theory and also better simulation results than conventional and other neural network methods.
Keywords :
image resolution; image restoration; interpolation; inverse problems; learning (artificial intelligence); least mean squares methods; neural nets; smoothing methods; LMS training algorithm; down-sampling; forward mapping; ill-posed problem; image superresolution restoration; interpolation algorithm; linear restoration; low-pass filtering; low-resolution image; neural network; residual errors; Computational modeling; Filtering; Image resolution; Image restoration; Interpolation; Low pass filters; Medical simulation; Multi-layer neural network; Neural networks; Nonlinear filters;
Conference_Titel :
Neural Information Processing, 2002. ICONIP '02. Proceedings of the 9th International Conference on
Print_ISBN :
981-04-7524-1
DOI :
10.1109/ICONIP.2002.1201927
