Title :
Mapping vector accumulator: fractal domain feature for character recognition
Author :
Linnell, T.A. ; Deravi, F.
Author_Institution :
Dept. of Electron., Univ. of Kent, Canterbury, UK
Abstract :
A new feature is presented which is calculated directly from the fractal-compressed form of images. The performance of this feature is analysed by using it in the training of standard classifiers for a character recognition problem. The feature is shown to be translation invariant, giving it an advantage over methods that use pixels or compressed data directly.
Keywords :
character recognition; data compression; feature extraction; fractals; image classification; image coding; character recognition; feature extraction; fractal domain feature; fractal image compression; mapping vector accumulator; translation invariance;
Journal_Title :
Electronics Letters
DOI :
10.1049/el:20046478
