DocumentCode :
394414
Title :
Distogram: a translation- and rotation-invariant and scale-covariant signature of a primitive shape
Author :
Lee, Geehyrik ; Sohn, Misook
Author_Institution :
Inf. & Commun. Univ., Seoul, South Korea
Volume :
4
fYear :
2002
fDate :
18-22 Nov. 2002
Firstpage :
1847
Abstract :
We propose a new one-dimensional signature of a two-dimensional binary pattern, which we call distogram meaning the histogram of distances between pairs of foreground pixels. A simple test consisting of six roman capital letters is carried out in order to evaluate the discrimination ability of the proposed signature. Also, a pulse-coupled neural network is described as an example of a simple network design that can compute a distogram given a two dimensional binary pattern as an input.
Keywords :
image recognition; neural nets; 2D binary pattern; distance histogram; distogram; primitive shape; pulse-coupled neural network; roman capital letters; rotation-invariant signature; scale-covariant signature; translation-invariant signature; two-dimensional binary pattern; Artificial neural networks; Biological system modeling; Computer networks; Computer vision; Frequency; Gray-scale; Histograms; Neural networks; Shape; Testing;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Neural Information Processing, 2002. ICONIP '02. Proceedings of the 9th International Conference on
Print_ISBN :
981-04-7524-1
Type :
conf
DOI :
10.1109/ICONIP.2002.1198993
Filename :
1198993
Link To Document :
بازگشت