A hybrid method for laser etched metal label recognition

In order to recognize the characters etched on the metal label, the structural properties of alphabetic and digital characters on the metal label are analyzed, a hybrid method is presented, and different features are used during the recognition. The end points, the three cross points, four cross poi...

Full description

Saved in:
Bibliographic Details
Main Authors: Huai-Bo Song, Chang-Hou Lu, Guo-Liang Lu
Format: Conference Proceeding
Language:English
Subjects:
Character recognition
Etching
four crossing points
Image segmentation
Information analysis
Pattern analysis
Pattern recognition
Pixel
structural feature
Text analysis
three crossing points
Wavelet analysis
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In order to recognize the characters etched on the metal label, the structural properties of alphabetic and digital characters on the metal label are analyzed, a hybrid method is presented, and different features are used during the recognition. The end points, the three cross points, four cross points and the positions of the three cross points and the end points are extracted. The templates made of those positions are used for the coarse classification. Minutiae features such as the Euler numbers are used for precise classification. It is demonstrated that the proposed structural features have good discrimination ability of similar characters and can be used for fast classification and recognition of etched characters on metal label.
ISSN:2158-5695
DOI:10.1109/ICWAPR.2007.4421606