A spatially-adaptive neural network approach to regularized image restoration

Palmer, Alex S., Razaz, Moe and Mandic, Danilo P. (2003) A spatially-adaptive neural network approach to regularized image restoration. Journal of Intelligent and Fuzzy Systems, 13 (2-4). pp. 177-185. ISSN 1064-1246

Full text not available from this repository.


When using a regularized approach for image restoration there is always a compromise between image sharpness and noise suppression. Since noise is removed at the cost of edges and detail within the image, there is a need to introduce algorithms which exhibit some kind of memory and cater for the spatial structure of an image. To this cause, we introduce an efficient restoration algorithm, based on a modified adaptive Hopfield neural network. The algorithm is capable of spatially regularizing an image and thereby preserving data fidelity around edges while simultaneously suppressing noise in more noticeable areas such as smooth regions. The proposed method demonstrates an improvement in restoration quality over existing adaptive and non-adaptive approaches. This is illustrated with simulations on benchmark images under varying noise levels.

Item Type: Article
Faculty \ School: Faculty of Science > School of Computing Sciences
Related URLs:
Depositing User: Vishal Gautam
Date Deposited: 04 Mar 2011 14:51
Last Modified: 15 Aug 2023 12:30
URI: https://ueaeprints.uea.ac.uk/id/eprint/21693

Actions (login required)

View Item View Item