University of Bologna

Dipartimento di Elettronica, Informatica e Sistemistica

Publication Database Query

Searched for "cm93:Kov"
Clicking on the number of each publication you can retrieve the bibtex entry. If available, the abstract can be accessed clicking on the title of the work.

Nearest-Neighbors Classification of Hand-Written Digits Using the CM-2

Zs. M. Kovacs-V. and R. Guerrieri


A statistical classifier for hand-written character recognition is presented. After a standard preprocessing phase for image binarization and normalization, a distance transform is applied to the normalized image, converting a black and white (B/W) picture into a gray scale one. The latter is used as feature space for a k-Nearest-Neighbor classifier, based on a dissimilarity measure which generalizes the use of the distance transform itself. The classifier has been implemented on a massively parallel processor, the Connection Machine CM-2. Results obtained from the classification of digits extracted from the U.S. Post Office ZIP code database are provided. The system has an accuracy of 96.73% on the digits when no rejection is allowed and it has an accuracy of 98.96% at 1% error rate.

Microelectronics Group Home Page Staff
Research Activities Teaching Activity
Publications Local Stuff
DEIS Home Page University of Bologna Home Page

External link, Local link, For local sites only, Stiil empty.