A KNOWLEDGE-BASED APPROACH FOR SCRIPT RECOGNITION WITHOUT TRAINING

被引：0
|
作者：
RAO, PVS
机构：
[1] Computer Systems and Communications Group, Tata Institute of Fundamental Research
来源：
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE | 1995年 / 17卷 / 12期
关键词：
CHARACTER SYNTHESIS; CURSIVE SCRIPT; DECODING; ENCODING; FEATURE MATRICES; ONLINE RECOGNITION; SCRIPT RECOGNITION; SHAPE VECTORS; TRANSFER FUNCTION; TRANSITION SEGMENTS;
D O I：
10.1109/34.476518
中图分类号：
TP18 [人工智能理论];
学科分类号：
081104 ; 0812 ; 0835 ; 1405 ;
摘要：
The approach is based on an empirical parametric model for the writing hand system. The parameters are so chosen and quantized as to retain only broad shape information ignoring writer-dependent and other variability. Concatenation of character prototypes generates archetypal reference words for recognition; training is unnecessary. Recognition scores exceed 90%.
引用
收藏
页码：1233 / 1239
页数：7
相关论文
共 50 条
12345