Loading...
Neuro-fuzzy profile clustering in image enhancement
Author(s)
Ngernplubpla, Jaturon
Date Issued
March 1, 2019
Type
Conference Paper
Abstract
This paper proposes a technique for clustering features into profile groups to obtain optimum enhancement weights for reconstructing high resolution images. Neuro-fuzzy model, which combines the fuzzy reasoning behavior with the adaptive learning capability and connectionist structure of neural networks, is adopted to analyze and learn with gradient data and statistics and to generate gradient profile priors. In enhancement process, the optimum weights are appropriately chosen according to the gradient profile priors. From the experimental results, the proposed algorithm demonstrates quantitative performance improvement in classifying data and perceptual improvement in recovery of the high-resolution image.
Citation
Ieecon 2019 7th International Electrical Engineering Congress Proceedings, 2019