DocumentCode
3598876
Title
Modular neural net architecture for one-step thinning algorithm
Author
Lee, Jin-Ho ; Park, Dong-Sun
Author_Institution
Dept. of Inf. & Commun. Eng., Chonbuk Nat. Univ., Chonju, South Korea
Volume
4
fYear
1995
Firstpage
1719
Abstract
A one-step parallel thinning algorithm using a deletion operator with 25 reference pixels and its neural implementation are presented. The one-step algorithm is an improved version of the previous parallel thinning algorithms in terms of the speed of the thinning process and the quality of thinned results. An efficient 2-layer neural net implementation of the one-step thinning algorithm is presented. The network consists of a module trained with the backpropagation learning rule and modules with hand-crafted weights
Keywords
image processing; multilayer perceptrons; parallel algorithms; recurrent neural nets; 2-layer neural net; backpropagation learning rule; deletion operator; hand-crafted weight; modular neural net architecture; one-step parallel thinning algorithm; Algorithm design and analysis; Character recognition; Costs; Image analysis; Image processing; Image segmentation; Neural networks; Parallel algorithms; Skeleton; Structural shapes;
fLanguage
English
Publisher
ieee
Conference_Titel
Neural Networks, 1995. Proceedings., IEEE International Conference on
Print_ISBN
0-7803-2768-3
Type
conf
DOI
10.1109/ICNN.1995.488879
Filename
488879
Link To Document