Proceedings of the Fourth International Conference on Document Analysis and Recognition
DOI: 10.1109/icdar.1997.620648
|View full text |Cite
|
Sign up to set email alerts
|

Recognising letters in on-line handwriting using hierarchical fuzzy inference

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1

Citation Types

0
5
0

Publication Types

Select...
5

Relationship

2
3

Authors

Journals

citations
Cited by 6 publications
(5 citation statements)
references
References 7 publications
0
5
0
Order By: Relevance
“…Features such as word length, diacritical marks, ascenders, descenders, combined as/descenders and segments crossing the word's axis, as well as several tolerance factors, are used. The first method was based on the letter candidates produced by a hierarchical fuzzy inference method [107], and the known distribution of the width of those candidates achieved a reduction of 44% of the hypotheses with an error rate of 0.5% for a 4126word vocabulary. Using the number of possible axis crossings instead of letter candidates leads to a reduction of 30% with the same error rate.…”
Section: Word Shapementioning
confidence: 99%
“…Features such as word length, diacritical marks, ascenders, descenders, combined as/descenders and segments crossing the word's axis, as well as several tolerance factors, are used. The first method was based on the letter candidates produced by a hierarchical fuzzy inference method [107], and the known distribution of the width of those candidates achieved a reduction of 44% of the hypotheses with an error rate of 0.5% for a 4126word vocabulary. Using the number of possible axis crossings instead of letter candidates leads to a reduction of 30% with the same error rate.…”
Section: Word Shapementioning
confidence: 99%
“…The extraction of these ink-boxes from dynamic handwriting data is described in more detail elsewhere [13], as is a simplified extraction method from static handwriting images 3. The representing point is chosen with respect to the desired zone-line, e.g.…”
Section: Methodsmentioning
confidence: 99%
“…Contextual information provided by overall statistics and previously obtained guidelines is used to adjust both representative point and tolerance. In order to eliminate undue oscillation of the approximating cubic spline function [13] (denoted by g(x); spline for short), the spline's control points are obtained by resampling the representative points and their associated tolerances.…”
Section: Methodsmentioning
confidence: 99%
“…The ink-usage experts enforce mutual exclusion as described in [5] and evaluate how much of the word image is used by the candidates. Every ink-segment ps s can be used in one of four modes per candidate: core, shared, latent and unused.…”
Section: Letter Candidate Expertsmentioning
confidence: 99%
“…First, holistic methods are used to reduce the size of the dictionary under investigation in serial result combination. As a segmentational approach, a letter-graph is produced by the Hierarchical Fuzzy Inference (HFI) method [5], a forwardreasoning production rule system, which thereby also furnishes automatically obtained hypotheses of -usually overlapping -letter boundaries. The second stage of the WCC then evaluates validity and quality of the constructed words.…”
mentioning
confidence: 99%